Performance based robust design optimization of steel moment resisting frames

• Robust design optimization is applied to seismic design of SMRF.
• The proposed method can provide designs that are safe, economical and robust.
• UDR is an effective efficiency indicator in seismic design of SMRF.

The seismic design of steel-moment resisting frames is subjected to uncertainties originating from various sources including imprecisely known seismic load, inaccurate modeling assumptions, as well as uncertain material properties and connection behavior. These uncertainties must be considered in the structural design process to ensure that a safe design is achieved. Design codes based on reliability of performance are useful in providing safety margins for the performance objectives with quantifiable confidence levels considering various sources of uncertainties. In these design codes, although seismic demand is usually calculated with a suite of ground motions, only the median seismic demand is used in the subsequent calculation of acceptance criterion (i.e., confidence level), and variation in seismic demand is not used. In the present manuscript, the authors utilize a performance based design approach for seismic design optimization of steel moment resisting frames, where material weight, mean value of seismic demand and variation of seismic demand are treated as three design objectives representing the cost, safety and robustness measures, respectively. Through a case study application, the proposed methodology is demonstrated to be capable of providing a set of Pareto-optimal designs with competing cost, safety and robustness. The obtained Pareto front designs are utilized in the development of uniformity drift ratio as design efficiency indicator. Required uniformity drift ratio to ensure efficient designs for each range of maximum inter-story drift is suggested based on the obtained results. Finally, the influence of the selected connection model and the response modification factor on the obtained results is investigated.