Published papers

In this study, we propose a method to optimize sections of a tensegrity structure considering large displacements during a large earthquake. Although there are multiple studies about the self-equilibrium of tensegrity structures, a method for their optimization or dynamic analysis that considers large displacements has not yet been proposed. Because a tensegrity system is a combination of struts and cables, i.e., compressive and tensile members, non-linear analysis is required to obtain realistic appropriations of displacements resulting from strong ground motions. We study the optimization of the minimum unit of a tensegrity structure and investigate the mode shape of the optimal design.

Detailed finite element (FE) analyses of a full-scale four-story steel frame structure, subjected to consecutive 60% and 100% excitations from the JR Takatori records during the 1995 Hyogoken-Nanbu earthquake, are conducted using E-Simulator. The four-story frame was tested at the largest shake-table facility in the world, E-Defense, in 2007. E-Simulator is a parallel FE analysis software package developed to accurately simulate structural behavior up to collapse by using a fine mesh of solid elements. To reduce computational time in consecutive dynamic time history analyses, static analysis with gravity force is introduced to terminate the vibration of the structure during the analysis of 60% excitation. An overall sway mechanism when subjected to 60% excitation and a story mechanism resulting from local buckling of the first-story columns when subjected to 100% excitation are simulated by using E-Simulator. The story drift response to the consecutive 60% and 100% excitations is slightly smaller than that for the single 100% excitation.

The concept of resilience is widely used in the disaster management field. In the earthquake-engineering field, many studies have investigated the structural performance of buildings and seismic resilience in terms of performance-based earthquake engineering. However, there are a limited number of studies on the social aspects of resilience. This study focuses on social activities by private companies. A company’s social activity is termed as its Corporate Social Responsibility (CSR). Further, this study also focuses on the relationships between the resilience of a community and the companies’ CSR activities in a local area. We conducted two questionnaire surveys for citizens and companies in Iwaki City, which suffered from the 2011 Tohoku earthquake. We hypothesized that social activities by companies would improve the resilience of the local community. In the questionnaire surveys, we set two types of respondents as follows: CSR staff of companies and citizens in the local areas. Using the questionnaire responses of the CSR staff, we analyzed characteristics of CSR activities using principal component analysis. Subsequently, using the local citizens’ responses, we compared the differences in their life recovery in terms of economic circumstances and social aspects.

This study focuses on verifying the effectiveness of the nonlinear power-law damping system on reduction of the vibration of base-isolated buildings with semi-actively implemented dampers over the ranges of resonance frequencies without causing detrimental effects over other frequency ranges. To verify the effectiveness of the proposed power-law damping system, shaking-table tests on a small-scale two-story physical building model are conducted. Random ground motions are used as the input excitations. The physical building model is equipped with a semi-active oil damper whose damping coefficient can be varied over four different values. This provides an effective mechanism for semi-actively implementing the power-law damping system. The shaking-table tests have demonstrated the effectiveness and robustness of the power-law damping and indicated that this can be a more practical solution to improving the performance of building base-isolation systems.

Control methods for a semi-active damper using Electro-Adhesive Gel (EAG) are analysed. Modelling of an EAG damper is conducted based on approximation of the force-voltage relation obtained from experimental data. The performance of an EAG damper is evaluated by shaking-table tests of a two-story small-scale specimen. The specimen is designed to adjust the ratio between first and second natural periods to that of a target real-scale base-isolated building, and system parameters are identified based on vibration response under white noise excitation. An EAG damper is implemented to the base-isolation layer of the small-scale specimen. Acceleration sensors are attached to each layer to acquire response data. As an observer in a control system, a Kalman filter is employed, and for a controller, LQG strategy and disturbance accommodating control are advocated to apply the suitable input voltage to the EAG damper. The cost function of the LQG is formulated to suppress the acceleration response of the specimen. Experiments using other control methods, simulated nonlinear spring and simulated dashpot, are also conducted for comparison. With respect to the input excitation, scaled earthquake ground motion records are used. Based on the results of the shaking-table tests, we report the validity of the EAG damper and discuss about effective control methods.

To reduce the amount of earthquake damage, the effectiveness of cooperative control between two controllers (one for the building and another for the equipment) was verified using a shaking table test. We employed a Kalman filter and the linear quadratic Gaussian strategy to minimise the amount of damage under strong ground motions, and the two controllers share the quantities of the state variables of the building and equipment. A three-storey specimen, which comprise a building (the main system) with an active mass damper and a base-isolated floor (a subsystem) with a semi-active oil damper, represents a 15-storey building of a data centre on a real scale. As a result of the shaking table test, we confirmed that implementing cooperative control reduced the relative displacement between the main system and subsystem without increasing absolute acceleration.

A sliding mode control is one of control methods with superior robustness against structural parameter change and uncertainty of a model. However, a sliding mode control requires multiple control parameters and a design method of these parameters has not been established. In this study, we propose a design procedure to determine the control parameters so as to minimize the damage probability in a certain service period of a building. A semi-actively controlled base-isolated building is studied and the damage probability based on the proposed design procedure is compared with that based on a conventional procedure that determines parameters to minimize the response under design input ground motion with a certain level of intensity.

Finite element analysis is carried out for simultaneously investigating global displacement and local stress responses of laminated rubber bearings under earthquake excitation. Natural rubber bearings are attached at the base of a building frame. The frame members as well as the rubber bearings are discretized into hexahedral finite elements with more than 3 million degrees of freedom. The material properties of rubber are represented by the Ogden model, and the frame is assumed to remain in elastic range. It is shown that the incompressible and hardening properties of natural rubber can be accurately simulated, and the deformation and stress distribution of rubber bearings of a frame under seismic excitation are successfully evaluated through a finite element analysis.

We conducted a theoretical analysis of the peak factor, defined as the ratio of the response maximum to the response standard deviation of a single-degree-of-freedom system. We constructed a peak factor formula that uses the natural circular frequency, damping factor, and duration of ground motion, and thereby estimated the peak factor of each mode of a multi-degree-of-freedom system. Using these mode-dependent peak factors, we then modified the existing complex complete quadratic combination method to estimate the maximum acceleration of a non-classically damped model. The response of a mid-story isolated building model under white noise ground motion was analyzed, and it was confirmed that the reformed formula globally provides good estimates in most cases.

The authors propose a compensation method using the Kalman filter and sensor reliability parameters to assure the continuous performance of vibration control even if sensor failures occur. The effectiveness of the proposed method was verified by conducting a shaking table test using a three-story specimen with an active mass damper, which is a small-scale model of a 15-story building. As a result of the shaking table test with simulated sensor damage, it was confirmed that the responses of inter-story drift angle and absolute acceleration were reduced by the proposed compensation method in comparison with a case without the proposed method.

Structural vibration characteristics of a semi-active base-isolated building were investigated using seismic observation records including those of the 2011 Great East Japan earthquake (Tohoku earthquake). Three different types of analyses were conducted. First, we investigated the long-term changes in the natural frequencies and damping factors by using an ARX model and confirmed that the natural frequency of the superstructure decreased slightly after the main shock of the Tohoku earthquake. Second, we investigated short-term changes in the natural frequencies and damping factors during the main shock by using the N4SID method and observed different transition characteristics between the first and second modes. In the second mode, in which the superstructure response is most significant, the natural frequency changed depending on the response amplitude. In addition, at the beginning of the ground motion, the identified first natural frequency was high possibly as a result of sliding friction. Third, we compared the natural frequencies and damping factors between the conditions of a properly functional semi-active control system and a nonfunctional system, by using the records of the aftershocks of the Tohoku earthquake. However, we could not detect major differences because the response was probably influenced by sliding friction, which had a more significant effect on damping characteristics than did the semi-active dampers.

In order to improve policy and programs to promote retrofitting of houses in Japan, incentives and disincentives for seismic risk management by homeowners were studied by two approaches: a fault tree analysis (FTA) method and a questionnaire survey to homeowners. The result of the FTA revealed two common causes that hindered homeowners’ seismic risk management: disaster awareness problem and fear of dishonest contractors. The questionnaire survey identified both incentives and disincentives. It was observed that neighbors could prompt retrofitting statistically significantly, and there were three major disincentives to retrofitting: high retrofitting cost, low contractor credibility, and little engineering information. The current policy in Japan put emphasis on seismic diagnosis in comparison with those in the United States. However, based on the above-mentioned observations, it was suggested that planning and reviewing of retrofitting work, as well as management after retrofitting, should be assisted more comprehensively in order to promote retrofitting. In addition, more attention should be paid to risk communication to provide engineering information on retrofitting, to foster mutual trust between homeowners and contractors/engineers, and to encourage information exchange with neighbors.

Vulnerability functions of wooden houses, which use the results of seismic diagnosis and an acceleration response spectrum, are formulated using equivalent linear models. They employ a probability density function of seismic performance indices of wooden houses, which are separated into an index of the resistance force and other indices related to the degradation of resistance force. The input seismic motion is evaluated based on an acceleration response spectrum and a natural period is determined by the indices of resistance force and threshold of damage. The application results to the 1995 Hyogo-ken Nanbu earthquake are compared with the damage statistics at the five sites, which have seismic records. The estimated damage ratios coincide fairly well with the actual damage surveyed by local governments.

This report presents a research project of developing “E-Simulator: virtual shaking table”, which is 　a simulation software based on a finite element method to analyze non-linear seismic response of constructions, as a part of E-Defense project. The key features of E-Simulator are 1) extremely large-scale numerical computation that realizes the finest temporal and spatial resolution of the structure response and 2) simulation of collapsing processes of structures that is comparable with the 3D full-scale model earthquake tests of E-Defense. This report explains the objectives of the E-Simulator and its two features. It also introduces the current state of E-Simulator project which includes the earthquake dynamic collapse analysis of 31-story steel framed structure and particle discretization scheme FEM for the RC structure.

Accuracy of estimated seismic waves using a one-dimensional ground model is examined. Hiyoshi Area in Yokohama is focused, which has relatively complicated topography. First, two one-dimensional layered ground models are prepared: Model 1, a four-layered model based on a boring log and Model 2, a two-layered model based on airborne laser profiler and geological cross sections. Seismic records with peak ground acceleration 8 to 141 cm/s2 are used for analysis, which were observed by six seismic stations in the study area. Incident waves at bedrock level of each observation point are calculated by one-dimensional equivalent linear analysis. Two ground models are compared with respect to acceleration response spectra. Then, acceleration response spectra at bedrock level at each observation point are compared in regard to each earthquake event. The results suggest that it is necessary to increase the spatial density of bedrock data to make an accurate ground model and to consider influence of surface waves and oblique incidence in regard to the studied area.

In order to develop damage detection method for a reinforced concrete (RC) structure using an impact hammer excitation, some indices based on Fourier spectrum of a response accelerogram were analyzed. Two test pieces of a RC beam were used: one has sufficient shear reinforcement and the other not. To cause nine stages of damage, a test piece was simply supported and vertical downward forces were loaded. At each damage stage, an impulse excitation was applied and the response acceleration was measured. It was revealed that an index using phase difference spectrum showed fairly good correlation with damage by which bending stiffness was reduced although the index used no information on excitation force. Between the two test pieces, clear difference of the correlation was not observed possibly because bending cracks proceeded for both test pieces.

Recently, buildings with an isolation device installed in middle story have been constructed. However, an adequate isolation layer level in height of a mid-story isolated building has not been established, and this causes a restriction on design of a plan of a mid-story isolated building. To reduce this restriction, semi-active controlled systems are compared to observe response characteristics. Regarding a control law, velocity feedback based on bilinear optimal control is used. Based on time history analysis with a 30-DOF system, response differences with respect to a position of an isolation layer is studied. It is observed that semi-active control scheme can be used to reduce the absolute acceleration of a roof and the deformation of an isolation layer. On the other hand, much attention must be paid to behavior of a lower structure under seismic excitation in design.

It is important to establish an accurate vulnerability function to estimate seismic damage. The purpose of this study is to expand vulnerability functions of wooden buildings using seismic diagnosis data based on equivalent linear models. The vulnerability functions consider characteristics of the region such as ground and building characteristics. The paper describes a method to calculate ratio of building damage in a region. The accuracy of the method is verified using the damage data of the Great Hanshin-Awaji Earthquake Disaster.

The questionnaire survey was conducted to clarify effective explanation of grades of seismic capacity using probability of damage. The panels living in Yokohama City, Kawasaki City or 23 Wards of Tokyo who were in their thirties to forties, and who wanted home ownership were selected and 759 valid answers were obtained. Questions regarding seismic risk perception and preference of seismic capacity of a house were asked. As a result, it was observed that the panels overrated seismic risk and demanded various seismic capacities. Additionally, it was revealed that explanation using the probability of damage in relation to seismic intensity was effective, and that a house with high seismic capacity was preferred even after information on risk of seismic damage, which might remove the bias of the overrated seismic risk, and on additional cost to improve seismic capacity were provided.

Due to past several earthquakes, damage of elevators, especially in rope, were reported. It is quite important to maintain the function of an elevator for rescue activity after a severe earthquake. This study clarifies the correlation between vibration of a main rope of an elevator and the characteristics of a building (stiffness and damping factor). Influence of the setting position of a driving machine is also investigated. A model of a thirty-story building with elevator system is analyzed on vibration modes and dynamics responses to several seismic ground motions. It is shown that two-elevator system, which sets a transfer floor in the middle height of a building, can reduce the response of a main rope, and that increased damping factor can also reduce the response.