Probability-based optimal design of friction-based seismic isolation devices

Abstract

In earthquake engineering the design of seismic isolation devices plays an important role to ensure structural safety and integrity. Such devices on one hand must allow for a sufficiently high level of structural decoupling and energy dissipation in order to reduce structural damage, and on the other hand must provide enough stiffness in order to prevent excessive deformations or residual offsets. This leads to trade-off considerations which can be dealt with through an optimization process. In the present paper, the earthquake excitation is treated as a non-stationary random process. Therefore, the design of the isolation device is based on a suitable probabilistic characterization of the dynamic response, i.e. the first-passage probability of critical response levels. The required first-passage probabilities are computed using a novel efficient Monte Carlo based simulation technique called asymptotic sampling. The design space is covered by using a design of experiment based on Latin Hypercube sampling. Due to the inherent statistical error of Monte-Carlo based analysis it is useful to apply a response surface technique (smoothing) based on the Moving Least Squares method. The design optimization involves conflicting objectives which can be resolved by applying a Pareto-type optimization approach. The main contribution of this paper lies in the connection of high-dimensional Monte-Carlo-based reliability analysis with multi-objective design optimization using a response surface approach and its application to an earthquake engineering problem.