Jordan Journal of Civil Engineering

Evaluation of Seismic Performance of PBD Optimized Steel Moment Frames by Means of Neural Network


Masood Danesh;


Performance-based design (PBD) is a method for structural design that accommodates different levels of structure performance against appropriate levels of earthquake intensities. On the contrary, reliability-based design (RBD) puts the reliability of the designed structure to achieve the desired performance levels as the basis of action and, unlike PBD, which utilizes deterministic design parameters, is of probabilistic geometrical and mechanical characteristics so that each of these probabilistic variables, with a specific mean and standard deviation, has a particular statistical distribution. In the present paper, 3, 6 and 12-story moment-resisting steel frames with special ductility are first optimized by using meta-heuristic optimization algorithms of PSO, ECBO, FA as well as finite difference algorithm (FDA), proposed by the author, based on the structure performance with initial cost objective function and using neural network. Thereafter, the reliability and collapse margin ratio (CMR) of the optimized frames are evaluated using the FEMA-P695 code. The obtained results show that although in terms of targeting the initial cost, the optimized frames have a structural weight nearly as low as possible; however, because of the relatively small reliability and CMR resulting from the use of deterministic design parameters by the PBD method, the frames will have a higher lifecycle cost. Therefore, in order to optimize such frames using PBD method, the use of the initial cost objective function is not recommended unless an optimal CMR value is included in the optimization constraints for their seismic performance or the optimization is performed by the RBD approach.


Structural optimization, Initial cost, Collapse margin ratio (CMR), Performance-based design, Reliability-based design, Neural network