Optimization of life-cycle cost of retrofitting school buildings under seismic risk using evolutionary support vector machine

Abstract

The assessment of the seismic performance of existing school buildings is especially important in seismic-disaster mitigation planning. Utilizing a support vector machine coupled with a fast messy genetic algorithm, this study developed two inference models, both using the same input variables: i.e., 18 building characteristics selected based on expert opinion. The first model was designed to judge whether a building needs to be retrofitted; and the second, to estimate the cost of retrofitting buildings to specific levels. The study proposes a life-cycle seismic risk framework that takes into account projections of the seismic risk a given building will confront over the course of its entire existence, and thus helps determine the economically optimal level of retrofitting. The results of a case study indicate that the higher upfront cost of retrofitting that is required to reach higher seismic performance levels could, depending on the level of predicted seismic risk, be offset by lower repair costs in the long run. It is hoped that this research will serve as a basis for further studies of the assessment of the life-cycle seismic risk of school buildings, with the wider aim of arriving at an economically optimal building-retrofit policy.