Damage state exceedance probabilities from aftershocks

Abstract

As demonstrated in the Canterbury earthquake sequence, shaking from aftershocks can be strong enough to cause significant damage and building collapse. This was a very specific case where in many locations the shaking from aftershocks was stronger than from the main shock; however, one important consideration is how vulnerable a building may become to aftershock shaking following shaking from a main shock. In this report we present a method for post-earthquake risk assessment for collapse of buildings.

Building on earlier work by Luco and others (Luco, et al., 2004; Yeo 2005), we extend the framework that is commonly used for probabilistic damage assessments prior to the occurrence of an earthquake. We start with the standard risk integral, which combines forecast ground motion hazard for a location (eg, from the New Zealand National Seismic Hazard Model; NSHM) with building models that predict how much damage a given structure will experience from a specific level of ground shaking. By combining this information, a forecast of the level of building damage expected for a particular time frame can be calculated. Due to models such as the NSHM not taking account of shaking from aftershocks, we have used a method to include the potential for ground shaking from not only a main shock, but from possible aftershocks. Additionally, we have developed and included structural modelling that accounts for any possible damage from a main shock and then determines the potential for different levels of damage from subsequent shaking from aftershocks; the damage probabilities output from such modelling is typically referred to as building fragilities.

The two additional components we present in this work are, given a main shock: 1) forecast modelling of the possible aftershocks and their anticipated ground motions; and 2) building fragilities that include damage from the main shock and potential aftershocks. The aftershock modelling is handled with a model which accounts for families of aftershocks and forecasts the locations of expected aftershocks (ie, the STEP model). We calculate the ground motions of the aftershocks using appropriate ground motion prediction equations, such as McVerry (2006) or Bradley (2010), for New Zealand. For the building damage modelling, we develop 4- and 5-storey reinforced concrete frame computer models that represent a common building type in New Zealand. We then model the buildings using two levels of complexity: 1) in a simple approximation, the building is only allowed to move in one direction; and 2) in a more detailed model, the building is allowed to move in all directions. In both cases we model the response of the building to recorded ground motion observations for multiple main shocks and aftershocks. Using the results of this, we develop fragility curves that estimate the damage of the generic NZ structure from back-to-back shaking from a main shock and aftershock. When combined using the risk integral described above, we are then able to estimate the probability for building collapse given expected (or observed) main shocks and the potential for aftershocks. One key outcome is in a comparison with an equivalent US reinforced concrete frame building where significant variability from the NZ model is observed; this indicates that the use of building models from other countries for developing building fragilities may be misleading.