Estimation of seismic response of mass irregular building frames using artificial intelligence

The determination of fundamental time period is an essential step in the seismic design process. The seismic design forces are inversely proportional to the fundamental time period. Therefore, underestimation of fundamental time period will lead to overestimation of design forces and will lead to uneconomical design. Majority of the buildings contain irregularity in some form or other due to functional and aesthetic considerations. These buildings have experienced large failures during the previous earthquakes. This is due to ignorance of the irregularity aspect in formulating the seismic design philosophies. This research work aims to address these short-comings and propose new design philosophies. To achieve this purpose, 256 building models with different degrees of mass irregularity have been modeled and analyzed using E-Tabs software. The analysis results are tabulated to create a seismic response databank. Regression analysis has been conducted on the response databank to propose new equations to estimate the fundamental time period. The feed forward artificial neural network employing Levenberg - Marquadrt algorithm has been used to train the input and output data. Finally, the proposed equations are validated against the code proposed expressions to demonstrate their efficiency.