Influence of imperfections on the strength of unstiffened, fabricated, tubular beam-columns: Fid-Bau Portal

Influence of imperfections on the strength of unstiffened, fabricated, tubular beam-columns

Hu, S.Z. / Prion, H.G.L. / Birkemoe, P.C.

AbstractFabricated tubular steel members are used in offshore structures as primary and secondary members; typical applications in fixed structures incorporate large diameter tubes in truss-type assemblies. These unstiffened tubes are formed by rolling plate into cylindrical components which are then seam and girth welded to create the circular tubular shape. The members produced in this fashion have characteristic imperfections which are different from those found in smaller diameter seamed or seamless hollow structural and pipe sections. These imperfections and their influence on the structural behaviour of beam-columns are subject of the study and results presented here.Earlier experimental studies at the University of Toronto on the influence of fabrication parameters are the foundation of the results reported here. Extensive measurements of as-fabricated geometry provided information on actual shape and these measured geometries are used to establish the analytical models using a finite element program, ANSYS; this program permits the modelling of non-linear material and geometric effects. The ultimate and post-ultimate local buckling behaviour was studied in the analyses to the same extent as observed in experiments.The load-displacement behaviour of tubular members depends significantly on the tube geometry and material properties. The discretization of the tube into elements is shown to require careful consideration so that potential deformation modes are not missed. Bilinear kinematic hardening material based on coupon tests of the ‘as fabricated’ steel, in combination with geometric modelling techniques yielded a behaviour response with a close resemblance to the experimental observations and measurements. The relative importance of various parameters is discussed and direct comparisons of experimental and analytical results with current code resistance formulations are made. The specific effects of residual strains are not addressed in this paper.