Structural steel design using second-order inelastic analysis with strain limits

Abstract

Steel framed structures are affected, to greater or lesser extent, by (i) geometrical nonlinearity associated with the change in geometry of the structure under load and (ii) material nonlinearity related to the onset and spread of plasticity. In traditional design approaches, the design forces and moments within structural members are usually determined from simplified structural analyses (e.g. first or second-order elastic analysis), after which member design checks are performed to assess the strength and stability of the individual members. The extent to which the strength and deformation capacity of cross-sections is affected by local buckling is typically assessed through the concept of cross-section classification. For example, only compact (Class 1) cross-sections are considered to possess sufficient rotation capacity for plastic hinges to develop and for inelastic analysis methods to be used. This approach results in step-wise capacity predictions and is considered to be overly simplistic. Since the structural analysis of steel framed structures is typically performed using beam finite elements, which are unable to explicitly capture local buckling, a more sophisticated treatment of the available deformation capacity is required if inelastic analysis methods are to be used for all cross-section classes. A novel method of design by advanced inelastic analysis has recently been developed (Gardner et al., 2019a; Fieber et al., 2019a, 2018a, 2018b [1–4]), in which strain limits are employed to represent the effects of local buckling in beam finite element models and thereby control the spread of plasticity and level of force/moment redistribution within a structure. It is thus possible to use a consistent advanced analysis framework to design structures composed of cross-sections of any class. In the present paper, application of the proposed design method to continuous beams and planar frames is illustrated and assessed. Ultimate load capacity predictions are compared against results obtained from benchmark shell finite element models that explicitly capture local buckling and to conventional steel design. It is found that the proposed method predicts safe-sided ultimate capacities that are consistently more accurate than current design methods, particularly for structures benefiting from strain hardening and for structures composed of non-compact cross-sections where inelastic redistribution is typically ignored.