Design of recursive filters for random wave simulation

Abstract

This paper is concerned with algorithmic tools for the efficient time domain simulation of random wave force processes. Instead of the summation of sine waves approach this work is concerned with the filtering approach to wave simulation. The filtering approach affords the possibility of uninterrupted generation of random records of arbitrary duration, making it possible to accumulate statistical information that requires long runs. Two types of filters are required, for initial generation and for transmission. The transmission filter is derived by seeking the relevant impulse-response function, both directly from hydrodynamical theory and indirectly as a transform. The finite length approximate (FIR) filters for transmission are replaced by suitably designed recursive filters. The error committed by the latter is compared with the error of an FIR filter of the same length. This comparison shows that the recursive filter enjoys an order-of-magnitude advantage in the error measure.