Estimation of seismic-induced demands on column splices with a neural network model

The current seismic design specification (AISC 341-05) requires that column splices in moment frames, when not made using complete joint penetration (CJP) welds, be designed to develop the flexural strength of the smaller connected column and the shear demand associated with flexural hinging at the top and bottom of a spliced column at a given story. AISC 341-05 assumes that the beam-to-column connection would reach its critical limit state before the column splice does. Estimating seismic demands on column splices involves both ground motion and structural parameters, i.e., it is a high order nonlinear and complex problem. This study presents a Neural Network (NN) model to estimate the seismic demands on column splices in low-, medium-, and high-rise steel moment frames. Nine input parameters and 6 output parameters were used to construct the NN model. The effect of each input parameter on the output parameters (seismic demands on column splices and frame) was investigated through a sensitivity analysis based on the NN model.