Earthquake occurrence modeling for evaluating seismic risks to roadway systems

The results of the application of a variety of techniques, which included bootstrap sampling, the use of antithetic values, the use of Latin squares sampling, use of control functions, a compound Poisson approach, and importance sampling, were presented. It was found that extremely large reductions in the number of simulations needed could be achieved for the mean and confidence limits of the conditional loss distribution. For the unconditional, annual-loss distribution, the reduction of the number of simulations achieved through post-sampling techniques was only a multiplicative reduction factor of slightly above 3.2.