Evaluation of the Ground Motion Record Selection Method for Structure Groups in case of Generic Steel Moment-Resisting Frames

Considering the fact that the computational cost of Nonlinear Time History Analysis (NLTHA) has been reduced significantly, specially, during last decade, this type of analysis has been promoted among civil engineering society. On the other hand, basic shift from conventional design approach to the performance-based method highlights the essential need for the estimation of different engineering demand parameters (EDPs) with as highest reliability as possible. Focusing on the selected set of strong ground motions (SGMs) as an important source of uncertainty on the results, there are a variety of studies in the earthquake engineering literature the purpose of which is to introduce a standard scheme for efficient selection and preparation of appropriate SGMs as the input of NLTHA. A part of existing differences, most of the methods suffer from a common limitation that is the application of simplifying assumptions in their contextual framework that may not always be correct. For example, there are several structure-specific scaling and selection methods using an equal SDOF as the representative of the target structure that must be analyzed. Although the use of such simplifications is unavoidable, their effect on the reliability of the results estimated by performing NLTHA under proposed set of SGMs by the method must be evaluated.
In this paper, a recently proposed structure specific record selection method is investigated in terms of its ability to keep the efficiency in case of structures that may challenge the assumption of reducing a MDOF nonlinear system to an equal SDOF. The investigated method have proposed an a priori set of SGMRs selected from a commonly used general set, which is introduced for collapse assessment. The method first utilizes the statistical exploration of a collapse capacity database that is constructed by analyzing numerous SDOF systems each of which represents specific combination of structural features such as ductility and period. Then, by defining a quantitative similarity measure, the whole database is refined to find the optimum subset representing the general set fairly good for any predetermined structural characteristics.
For a comprehensive study, a group of 2-D one-bay vertically regular frames of five different heights (i.e. 3, 6, 9, 12, and 15 stories) has been used in this study. The height-wise distribution of stiffness was tuned to achieve equal drifts in all stories that are calculated using the Iranian code of practice for seismic resistant design of buildings forces (Standard 2800). The yield strength distribution was chosen such that the yielding is observed almost simultaneously at all plastic hinges under the lateral force distribution, which is determined by Standard 2800. To cover a variety of ductility values, nonlinear response spectra with constant ductility equal to 1, 2, 4, and 6 have been utilized as the design spectra. Thus, eight different designs (two values of T1 and four values of μ) are considered for each fixed height, leading to a total of 40 frames. A set of 22 pairs of horizontal SGMRs selected from a specific far-field set, which was used in the FEMA P. 695 as the suggested SGMs for NLTHA.
The most important results that can be concluded from the comprehensive evaluation of the mentioned selection method can be listed as;
• The proposed selected subsets, in most cases underestimate seismic demands of regular frames compared to the estimations by using the reference set of 44 SGMs. This can be attributed to the fact that many of steel frames do not experience severe levels of nonlinearity or side-sway collapse.
• The application of the proposed subsets by the selection method does not necessarily result in the reduction of the statistical dispersion in the estimated EDPs.
• Comparing the estimated EDPs by the proposed subset with those of other potential subsets, almost in all cases there is no significant superiority of the selection method in terms of accuracy and reliability. This can be interpreted by noting to the fact that the goal of selection method is the reliable collapse simulation of the structures and it cannot easily be generalized to the other performance levels.
The results confirm that the ductility plays an important role in the SGMs selection output. Therefore, it is suggested that the provision of a new selection process involving ductility of target structure is investigated as a future complementation of the current method.