DEVELOPING FUNDAMENTAL PERIOD FORMULAE FOR STEEL FRAMED STRUCTURES THROUGH MACHINE LEARNING AND AUTOMATED ALGORITHMS

Abstract

The fundamental period of a structure is an important intrinsic dynamic property of the structural system response. Current design codes use simplified empirical formulae to estimate the fundamental period, that generally do not account for the influence of plan geometry, cross-sectional properties of the specific structural elements, nor the properties of the soil the structures are founded on, and thus the effects of soil-structure interaction (SSI) are not accounted for. This research work investigates the fundamental period of unbraced steel framed structures of various geometric configurations, and multiple different steel column members, as well as the SSI effect. A new algorithm is also developed for the automated construction and analysis of finite element models. The proposed automated procedure is used to create a dataset that is used to train and test the predictive formulae for computing the fundamental period of steel structures with and without accounting for the SSI effect. The proposed 40-feature formula was found to derive an optimum coefficient of determination (R2) with a value of 99.976% and an error that is less than 2%.