Keywords: reliability, soil-structure interaction, performance-based design, Monte Carlo simulation, radial basis function, neural networks.

The modern approach to design of structures subjected to seismic loading is based on the principle that a structure should meet performance levels according to specified reliabilities over the service life. In fact, the purpose of this approach is to obtain designs with a more predictable seismic behavior, quantifying and controlling the risk to an acceptable level. Therefore the total cost would be minimized during the service life of the structure. The concept of performance-based design (PBD) has been introduced based on this approach by SEAOC Vision 2000 guidelines [1].

The problem of soil-structure interaction (SSI) is a topic of interest in engineering practice. This problem has a considerable influence on the response of massive structures such as dams, bridges, and multi-storey frames. Past researches have shown that the seismic responses of structures differ when they are founded on a deformable or stiff soil. The SSI may increase the natural periods of the structure and change the system damping arising from wave radiation [2].

In this paper, finite element modelling and seismic responses of SSI are combined to achieve a target annual failure probability for each of the specified performance levels. A nonlinear dynamic SSI system is used to obtain the structural response, implementing lumped plasticity for beam-column elements and elastic-plastic behaviour of soil with multi-yield-surface J₂ plasticity. The performance functions or limit states associated with each performance level are calculated by the structural response. Then the probability of non-performance and the annual probability of failure are evaluated corresponding to each of the limit states considered in a given performance level. For each of the performance levels, the probability of non-performance is estimated using MCS. To reduce the computational time for calculating the probability of non-performance, the structural response are evaluated using properly trained radial basis function neural networks [3]. Finally, the annual probabilities of failure are compared with target values of failure probabilities which are suggested for each performance level.

1

SEAOC Vision 2000 Committee, "Performance Based Seismic Engineering of Buildings", Structural Engineers Association of California, Sacramento, California, USA, 1995.

2

J.P. Wolf, "Dynamic Soil-Structure Interaction", Prentice-Hall, Englewood Cliffs, 1985. doi:10.1016/B978-075066164-5/50012-5

3

E. Salajegheh, S. Gholizadeh, M. Khatibinia, "Optimal Design of Structures for Earthquake Loads by a Hybrid RBF-BPSO Method", Journal of Earthquake Engineering and Engineering Vibration, 7(1), 14-24, 2008. doi:10.1007/s11803-008-0778-y

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £145 +P&P)