Influence of stochastic geometric imperfections on the load-carrying behaviour of thin-walled structures using constrained random fields

• Published in: Computational mechanics 2018-11, Vol.62 (5), p.1107-1125
• Main Authors: Lauterbach, S., Fina, M., Wagner, W.
• Format: Article
• Language: English
• Subjects: Classical and Continuum Physics; Computational Science and Engineering; Correlation analysis; Defects; Dependence; Empirical analysis; Engineering; Fields (mathematics); Finite element method; Intersections; Lower bounds; Mathematical models; Original Paper; Stability analysis; Stochastic processes; Structural engineering; Theoretical and Applied Mechanics; Thickness; Thin wall structures
• ISSN: 0178-7675; 1432-0924
• DOI: 10.1007/s00466-018-1554-0

Since structural engineering requires highly developed and optimized structures, the thickness dependency is one of the most controversially debated topics. This paper deals with stability analysis of lightweight thin structures combined with arbitrary geometrical imperfections. Generally known design guidelines only consider imperfections for simple shapes and loading, whereas for complex structures the lower-bound design philosophy still holds. Herein, uncertainties are considered with an empirical knockdown factor representing a lower bound of existing measurements. To fully understand and predict expected bearable loads, numerical investigations are essential, including geometrical imperfections. These are implemented into a stand-alone program code with a stochastic approach to compute random fields as geometric imperfections that are applied to nodes of the finite element mesh of selected structural examples. The stochastic approach uses the Karhunen–Loève expansion for the random field discretization. For this approach, the so-called correlation length l c controls the random field in a powerful way. This parameter has a major influence on the buckling shape, and also on the stability load. First, the impact of the correlation length is studied for simple structures. Second, since most structures for engineering devices are more complex and combined structures, these are intensively discussed with the focus on constrained random fields for e.g. flange–web-intersections. Specific constraints for those random fields are pointed out with regard to the finite element model. Further, geometrical imperfections vanish where the structure is supported.