Analysis of the efficiency of obtaining reliable data when operating with number sets in simulation of bridges structural components

Abstract

Abstract The issue of reducing the complexity of a multifactorial simulation experiment to an acceptable level is relevant. The simulation complexity is characterized by the computing capabilities of a PC while maintaining a given level of accuracy and certainty of the problem being solved. Many authors neglect the justification of the sample size, thereby reducing the experiment’s reliability. Loads are stochastic, therefore, in the calculations, they are determined by rated and design values. However, the Monte Carlo methods allow the specific realization of loads based on the distribution law chosen. Techniques for optimizing the multifactorial simulation experiment procedures have been proposed on a test example of adding permanent loads due to the bridge pavement layers. They allow excluding the procedure of building multidimensional laws of the final distribution of random parameters, while sequentially transforming them from input to output values. The minimum required simulation experiment size, which should exceed 7.5 106 realizations and is determined by the product of the number of tests N by the number of runs n, has been experimentally justified for the structural components of bridges.