Probabilistic Methods in Optimization of Structures Under Earhtquake Excitation


Abstract eng:
The process of analysis and design of structures under earthquake excitations depends on many random variables, reflecting the uncertainties in the mechanical properties, dimensions of structural components and, most important, in the characteristics of the ground motions. It is necessary to take into account all these uncertainties. The process of performance-based design then calls for an optimization, usually for minimum total cost, under minimum reliability constraints for the different performance requirements. This process can be organized in three main blocks: a) A nonlinear dynamic analysis is used to obtain structural responses for each combination in a deterministic set of the random variables, and for a set of ground motions likely to occur at the site. These sets are specified with the help of experimental design, and the ground motions could be either obtained from historical data or be artificially generated. The analysis is performed for each ground motion and each variable combination, and the results stored in databases. These, in turn, are used to train neural networks which permit the approximate evaluation of the structural responses, avoiding further dynamic analyses. b) The reliability level achieved for different combinations of the design parameters (usually, in the case of reinforced concrete, the means of structural dimensions or steel reinforcement ratios) are then calculated. These reliability levels are obtained by simulation, using the neural networks for structural responses. The reliability levels achieved for each performance level are also represented by neural networks, with the design parameters as inputs. c) The optimization involves the minimization of a total cost objective. The total cost, or life-cycle cost, includes the original associated with construction, plus the one associated with repairs due to damage produced by future earthquakes during the structure’s economic life. The earthquakes are assumed to arrive randomly during the structural lifetime. Also the human costs of injuries or casualties during an earthquake are included, plus those for temporary rent during repair. The paper describes an effective optimization algorithm base on a random search, avoiding the calculation of gradients. The results provide the set of design parameters which minimize the total cost, while satisfying minimum target reliability levels for each of the performance criteria. As an example, the paper includes optimization results for a four-story reinforced-concrete office building with different attributes: (i) a basic structure with portal frames; (ii) the same structure but including energy dissipation devices based on steel yielding; (iii) the same basic structure but introducing lead rubber bearing type base isolators. The results compare the total costs for each case, the corresponding optimal design parameters and the achieved reliabilities when minimum targets are not imposed.

Contributors:
Conference Title:
Conference Title:
16th World Conference on Earthquake Engineering
Conference Venue:
Santiago (CL)
Conference Dates:
2017-01-09 / 2017-01-13
Rights:
Text je chráněný podle autorského zákona č. 121/2000 Sb.



Record appears in:



 Record created 2017-01-18, last modified 2017-01-18


Original version of the author's contribution as presented on USB, paper 1674.:
Download fulltext
PDF

Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)