Deterministic versus probabilistic seismic hazard analysis for critical structures

Probabilistic seismic hazard analysis has been practically unchallenged since its inception three decades ago. However, information has been accumulating which shows convincingly that PSHA is a defective procedure. Its greatest weakness is the dependence of the probability theory on the Gutenberg-Richter magnitude and a recurrence relation which can no longer be regarded as a power law. Remedies that rely on incorporating paleoseismic information and characteristic earthquakes into the probability calculation introduce other errors resulting from fragmentary data and the known non-uniformity of earthquake occurrence in space and time. The worst corrective for probability is the method developed by the Electric Power Research Institute and the Lawrence Livermore National Laboratory that averages multiple expert opinions. Expert opinions cannot be averaged meaningfully because the criteria for different models are nonequivalent. On the other hand, the deterministic procedure for earthquake hazard evaluation avoids the above defects by eliminating the falsely precise time element in the probabilistic estimation. Geologic time for recurrence is used, according to accepted criteria such as a single movement in the past 12,000 years or multiple movements in 500,000 years. For a critical project, where the consequences of failure are intolerable and protection is needed against the worst that can be reasonably expected to occur (the maximum credible earthquake), the deterministic method is strongly recommended.