This study presents a comparison of a hybrid approach (Hyb) and a fully probabilistic approach for incorporating site effects in a probabilistic seismic hazard analysis (PSHA). In the employed Hyb, the amplification factors (AFs) are computed by performing equivalent‐linear site‐response analyses. The input rock motions are derived for two separate earthquake scenarios, controlling the low‐ and the high‐frequency portion of the rock ground motions based on the deaggregation results. The rock ground motions are then multiplied by the computed AFs to define the surface ground motions. The considered fully probabilistic approach is characterized by higher levels of sophistication in which the rock hazard is convolved with the probability density function of the amplification functions to calculate the surface ground motions. In both employed approaches, the uncertainties associated with the site characterization are addressed via Monte Carlo randomization.
The employed approaches are highlighted by performing a site‐specific PSHA for a liquefied natural gas (LNG) tank located in the Gulf Coast region. The seismic analysis of an LNG tank is a challenging and complicated task because a wide range of frequencies should be considered, covering both the fundamental frequency of the structure (between 2 and 10 Hz) and a sloshing frequency of around 0.1 Hz. Furthermore, the considered site is located in the Gulf Coast region where the thick sediments exhibit significantly different ground‐motion attenuation in the region. A pressing concern in the evaluation of seismic hazard in this region is using the appropriate attenuation relationships, which are not only tuned to the unique site condition of the region, but are also valid for longer periods required in LNG tank seismic design. The employed approaches resulted in very similar surface ground motions, which reveals that increasing the level of sophistication in the numerical ground response analyses might not be always required.