Computational Improvement for Expected Sliding Distance of a Caisson-Type Breakwater by Introduction of a Doubly-Truncated Normal Distribution

Abstract

The objectives of this paper are to investigate the validity of conventional evaluation methods for expected sliding distance of a caisson-type breakwater, and to propose alternatives for more accurate computation of the expected sliding distance. In particular, the method for considering uncertain factors, which appear in the design process, and the influence due to the seed of random variable on the expected sliding distance are discussed. The present paper shows that simulated result under the original normal distribution causes abnormal sliding distance because some random variables are taken outside the region where experimental or observed values are valid. It also points out that the seed of random variable largely affects the expected sliding distance in the computation process. Therefore, we have employed a doubly-truncated normal distribution instead of the original one for accurate computation of sliding distance, and we have introduced the concept of average expected sliding distance as the mean value of expected ones computed for various different seeds. The present paper has proven that the employment of the doubly-truncated normal distribution and the concept of average expected sliding distance are very effective on accurate estimation of the expected sliding distance.