Full Session with Abstracts
Seismic design of frame buildings has been well investigated over the years. Frame structures are well known for their energy dissipation under seismic loads due to the development of plastic hinges. However, most of the optimal design procedures that were presented to this day displayed either demanding great computation power (in particular in large-scale problems) or leading to a non-practical design in regard to section properties in comparison to a standard profile table. In order to cope with these issues, this paper presents a new methodology for performance-based optimal practical seismic design for moment resisting steel frames. The goal of the optimization process is to find a minimum cost design that satisfies an inter-story drift constraint.
The problem formulation considers the section properties as continuous variables and are determined by a gradient-based optimization algorithm. Gradient-based optimization approaches are well known for their efficiency in terms of computational effort, especially in comparison to heuristic approaches (e.g. genetic algorithms). However, the real nature of the problem is discrete. For this reason, discrete material optimization (DMO) is applied in order to achieve a practical solution, in terms of section properties and their matching to a standard profile table. The DMO function is based on material interpolation techniques that were successfully applied in the context of discrete optimization problem (e.g. topology optimization).
The described optimization procedure is based on a nonlinear response history analysis that is used to evaluate the inter-story drifts while subjecting the structure to a realistic set of ground motions. For the gradients computation, the adjoint sensitivity analysis for nonlinear systems is used, in particular, the discretize-than-differentiate approach is adopted.
The paper offers an innovative optimization procedure for design professionals as well as researchers who deal with either structural optimization or earthquake engineering. For practical engineering, it offers a new robust tool for the design of new steel frame structures under seismic loads. The two main advantages of this procedure are expressed in the practical solution outcome and in the efficiency in terms of computer running time, because of this reason a large-scale problem can be easily solved by a regular PC as well. The proposed methodology can be a step toward a realistic minimum cost seismic design of steel frame structures.
Preliminary results showed a good convergence to a discrete design and supported the high-efficiency claim.