Loading Protocols for Structures Designed for Different Behaviour Factors

All the existing loading protocols have been developed independently of the value of the behaviour factor q that the structure has been designed for. Conservatively, a single loading protocol is adopted, for all q values, based on the structural system with the most onerous cumulative seismic demands. However, analyses conducted by the authors, show that imposed cumulative seismic demands decrease significantly for structures designed for high behaviour factors. This drives to the conclusion that adopting a single loading protocol for the most demanding q value may lead to the derivation of highly conservative loading protocols for the rest of the structural systems. In this paper, new loading protocols will be developed for structures designed for different behaviour factors. Introduction Performance-based earthquake design and assessment requires reliable estimates of structural members’ strength and deformation capacities. These capacities can often not be predicted accurately by analytical or numerical modelling and experimental testing is required. Most commonly, quasi-static cyclic tests are conducted where predefined displacement histories, named loading protocols, are applied at slow rates. When subjected to cyclic loading, strength and in particular deformation capacity of structural components depend on the imposed cumulative damage demand (Krawinkler et al. 2001). Hence, in order to yield realistic capacity estimates, loading protocols must reflect the estimated cumulative seismic demands on the structural components of interest. Several loading protocols have been developed in the literature for different types of structural and non-structural components. Examples are: SPD protocol (Porter 1987), CUREE protocols (Krawinkler et al. 2001), EN-12512 protocol (EN 2001), FEMA-461 protocols (FEMA 2007) and ISO protocol (ISO 2010). Recently, Mergos and Beyer (2014) developed a suite of loading protocols for different seismicity levels (low to moderate vs. high), hysteretic models, fundamental periods and number of cycles per load step (one, two or three). All loading protocols follow the same analytical form which requires only two parameters to define the amplitudes of each load step. Adopting this approach, instead of proposing a single protocol, provides more representative and less conservative loading protocols for the different structural systems and levels of seismicity. All the existing loading protocols, including the ones proposed previously by the authors (Mergos and Beyer 2014) have been developed independently of the value of the behaviour factor q that the structure has been designed for. Conservatively, a single loading protocol is adopted, for all q values, based on the structural system with the most onerous cumulative seismic demands. Figure 1, presents the variation of the number of damaging cycles N and the sum of normalized, with respect to the maximum, cycle amplitudes Σδi with the behaviour factor q (Mergos and Beyer 2014). The results were obtained by performing 45360 time history analyses on 567 SDOF systems with different behaviour factors, periods of vibration, hysteretic models and hardening ratios. It can be seen that N and Σδi vary significantly and tend to decrease as q increases. This drives to the conclusion that adopting a single loading protocol for the most demanding q value may lead to the derivation of highly conservative loading protocols for the rest of the 1 Lecturer, City University, London, [email protected] 2 Assistant Professor, EPFL, Lausanne, [email protected] P MERGOS and K BEYER 2 structural systems. This observation becomes more important when considering that the higher cumulative demands occur, typically, to structural systems designed for small behaviour factors. Cumulative demands of these systems are developed at small levels of ductility demands. Applying the same cumulative demands to structural systems that develop significant ductility demands becomes even more conservative. In this paper, new loading protocols will be developed for structures designed for different behaviour factors. This will be done by applying the same methodology as described in Mergos and Beyer (2014). It is anticipated that the new loading protocols will reflect better cumulative damage demands of structural systems driving to more realistic estimates of their structural capacities from experiments. Figure 1. Variation of cumulative damage demands with the behaviour factor: a) Sum of normalized cycle amplitudes Σδi ; b) Number of damaging cycles [Mergos and Beyer 2014] Selection of ground motions and structural systems Cumulative seismic demands depend strongly on the selected ground motion records and hysteretic behaviour of structural systems. The previous drive to the conclusion that the selected ground motion records and hysteretic models should be representative of the level of seismic hazard and the type of structural system respectively the loading protocols are developed for. In order to establish by means of quasi-static cyclic testing reliable estimates of the deformation capacity at the near collapse limit state ΔNC in accordance with EC8 design objectives, the proposed loading protocols should represent the 2/50 seismic hazard level. For this reason, selection and scaling of the ground motion records in this study aim at representing the cumulative demand imposed by this seismic hazard level. Furthermore, two different levels of seismicity are examined herein. This is the case because previous analyses conducted by the authors of this study (Mergos and Beyer 2014) have shown that the level of seismicity affects importantly cumulative damage effects imposed to structural systems. For high seismicity regions, 20 ground motion records, used in several previous similar studies (e.g. Krawinkler et al. 2001, FEMA-461 2007), are employed. For low to moderate seismicity regions, a set of 60 ground motion records representative of the 2/50 seismic hazard level of the city of Sion in Switzerland is used. More information regarding this ground motion set can be found in Mergos and Beyer (2014). The selected ground motion records are scaled one by one in order to match the spectral acceleration of the horizontal elastic spectrum of EC8 for the 2/50 seismic hazard level at the fundamental period of the structure. The target EC8 elastic spectrum is derived for soil class C. 0 1 2 3 4 5 6 0 2 4 6 8