Abstract
A design method based on interval estimation for determining the appropriate number of samples for seismic analysis using the Monte Carlo method was developed. The Monte Carlo method requires an appropriate number of samples to produce valid results. However, an increased number of samples might increase in computing cost. It is therefore desirable to determine a reasonable number to give the required accuracy. The developed method with use of interval estimation can determine the relationship between number of samples and the error for mean and standard deviation, which are the dominant parameters that define a probability distribution of seismic responses. Accordingly, the number of samples needed to produce the results with required accuracy can be determined from the relationship. The results of the developed method were compared with those of the Monte Carlo method for elastic and inelastic single-degrees-of-freedom systems. It is demonstrated that the developed method can determine the reasonable number of samples needed to produce the mean and standard deviation of seismic responses with the required accuracy.
