An Efficient Methodology that Simulates a Multi-Dimensional Non-Gaussian Field to Evaluate the Effect of the Spatial Distribution of Corrosion in a Steel Beam

This paper presents a methodology to simulate multi-dimensional non-Gaussian random fields and uses this to model, in a probabilistic sense, the spatial distribution of corrosion in a steel beam. In particular, the proposed approach is an extension of a recently developed iterative technique that generates one-dimensional, uni-variate sample functions of non-Gaussian random fields and processes. The contribution of this work is the extension to the case of multi-dimensional random functions and its application to a steel cantilever beam with simulation of corrosion penetration. The amount of corrosion is modeled as a random field and applied to a finite element mesh on the top flange. The field matches both the arbitrarily prescribed spectral density function and the non-Gaussian marginal distribution. The effect of the spatial variability of the corrosion on the flange is evaluated using the tip displacement as representative structural response metric, while a series of finite element analyses are performed for a Monte Carlo simulation. The results show that the spatial variability has a substantial effect on the structural response and an equivalent deterministic model with the same average corrosion uniformly distributed would not capture properly the structural behavior.

Language

  • English

Media Info

  • Media Type: Web
  • Features: References;
  • Pagination: pp 1059-1069
  • Monograph Title: Structures Congress 2014

Subject/Index Terms

Filing Info

  • Accession Number: 01522520
  • Record Type: Publication
  • ISBN: 9780784413357
  • Files: TRIS, ASCE
  • Created Date: Apr 9 2014 3:02PM