Bias assessment in Incremental Dynamic Analysis due to record scaling

Incremental Dynamic Analysis (IDA) involves a series of nonlinear response history analyses with a suite of incrementally scaled ground motion records. Although IDA is perhaps the most comprehensive seismic performance assessment method, it receives criticism because several ground motion records are scaled up until the structure collapses. The scaling practice often results to unrealistic multipliers, thus modifying the amplitude of the ground motion and introducing bias on the structural performance estimation. Record scaling is a common practice in earthquake engineering due to the lack of natural records corresponding to large magnitudes and/or small distances from the fault rupture location. In this work we use a large number of ground motion records to compare the predictions of IDA with that of unscaled ground motions and we propose a new methodology in order to quantify the bias introduced in IDA. Apart from natural records, we have conducted broadband ground motion simulations for rupture scenarios of weak, medium and large magnitude events in order to expand our record database. The investigation is performed on a series of inelastic single-degree-of-freedom systems and on two multistory steel moment frame buildings. The results pinpoint both qualitatively and quantitatively, for the full range of limit-states, the bias that IDA introduces on the structural performance estimation.