ON ADVANCED MONTE-CARLO SIMULATION PROCEDURES IN STOCHASTIC STRUCTURAL DYNAMICS

As analytical methods to predict the response of MDOF-systems under st ochastic loading are quite limited, numerical procedures, such as Mont e Carlo simulation techniques, are frequently applied. Direct Monte Ca rlo simulation, however, particularly for reliability analyses, is not suitable for providing sufficient information on the tails of the dis tribution of the response. In this paper Monte Carlo methods, applicab le to Markov processes are further advanced, i.e. by leading the gener ated samples towards the low probability range which is practically no t accessible by direct Monte Carlo methods. In this context, notions a s developed for variance reduction techniques are introduced and discu ssed. Based on criteria denoting those realizations which lead most li kely to failure, a numerical procedure called ''Double & Clump'' (D&C) is discussed briefly. It is shown that the rather numerically involve d D&C procedure can be simplified by a technique known as ''Russian Ro ulette and Splitting'' (RR&S). In two numerical examples, both procedu res D&C and RR&S are compared with direct Monte Carlo simulation as we ll as the Response Surface Method (RSM) demonstrating comparable accur acy. (C) 1997 Elsevier Science Ltd.