Elastic Demand Spectra
Descripción
Elastic Demand Spectra Kyriazis Pitilakis, Anastasios Anastasiadis, Dimitris Pitilakis, Konstantinos Trevlopoulos, Konstantinos Senetakis
Aristotle University, Thessaloniki, 54124, Greece
Abstract Elastic demand spectra for performancebased design are normally estimated from the acceleration design spectra proposed in seismic codes, for different general soil categories under freefield soil conditions. Their accuracy must be perman ently checked using recently acquired strong ground motion records from stations with wellconstrained soil conditions. A comparison of the EC8 elastic response spectra for soil categories B and C, with more than 300 high quality worldwide re cords, is presented in the present paper. It is proved that the EC8 demand spectra need further improvement, in order to be better correlated with actual recordings. Moreover, the design demand spectra when applying the principles of the per formancebased design, should take into account the effects of the soilfoundation structure interaction and the potential foundation soil improvement, which may modify considerably the design input motion, compared to the freefield condi tions. The representative cases presented herein demonstrate the importance of these two issues in the aseismic design of structures. The final conclusion of the present illustrative work is that the elastic demand spectra to be used for an accur ate seismic design, may differ considerably from the demand spectra provided in the code, which do not take into account the issues that are shortly described and discussed in this paper.
Introduction Performancebased seismic design requires the estimation and selection of elastic acceleration and displacement spectra for different damping ratios in a wide range of periods. In this study, demand spectra from a well constraint database of strong ground motion records from Japan, Greece and United States, in well defined soil
2
conditions, are computed. The results are presented for different soil categories, and for two different seismic excitation levels, as it is suggested in EC8. The dis cussion is focused on the differences between the design demand spectra of EC8 and the computed demand spectra from a large data set of real records. Traditionaly, the demand spectra are calculated at the foundation level for the freefield soil response, assuming linear soil behavior. However this is not true in practice. The soil behavior is not linear elastic and the the soilfoundation structure interaction may modify considerably the design input motion and hence the elastic demand spectra. In this paper, an equivalent linear approximation is implemented in the substructure approach (D.Pitilakis and Clouteau 2009) of soil foundationstructure interaction (SFSI). The theoretical model is presented and then applied on a simple soilfoundationstructure (SFS) configuration. The simplified model is subjected to different input ground motions, aiming to excite the nonlinear soil behavior. The combined effects of the SFSI and the nonlinear soil behavior on the spectral demand of the system are highlighted through parametric analyses in the context of the performance based design. Finally, in many real cases, where important structures are founded on soft soils of low strength, it is necessary to improve soil conditions using different techniques. One of the most popular is the construction of stone columns. A re cently developed alternative technique replace at a given depth the initial soil with a compacted mixture of rubber and sand material (RSM) (Anastasiadis et al. 2009). The elastic demand spectra under these conditions may differ considerably from the demand spectra of the initial soil. The behavior of a simple SFS system founded on a soft cohesive soil that corresponds to soil category C according to Eurocode 8 replaced with rubbersand compacted layer, is examined in the present paper. Three RSM materials having 0%, 5% and 25% rubber by mixture weight ratio are used in this study. The overall depth of the replacement columns is 20m, and the ratio of face replacement is on the order of 10%. The effect of the soft soil RSM reinforcement on the demand spectra of a SDOF structure is depicted for the seismic input of the Athens, Greece 1999 earthquake. Similar analysis where per formed for the traditional stone column improvement. The aim of these three parts of the work is to validate the reliability of EC8 de mand spectra with a large sample of high quality worldwide records, and to high light some important aspects of the selection of the design elastic demand spectra in view of the performance based design. Real structures on good or poor soil con ditions are rarely subjected to the elastic demand spectra proposed by seismic codes, which correspond to an ideal freefield ground motion. SFSI, soil nonlin earity and soil improvement may modify considerably the design elastic demand spectra.
3
Comparison of EC8 elastic demand spectra with worldwide strong ground motion records
Selection of strong ground motion records A large set of high quality, mainly digital, records worldwide has been selected and processed in order to calculate separately the horizontal elastic acceleration and the displacement spectra directly from the respective time histories after prop er filtering. The results are then presented in terms of elastic demand spectra. A large number (~1000) of welldocumented strong ground motion recordings with different PGA values, Mw magnitudes and epicenter distances were selected for the soil classes studied in this paper (Fig.1). The selection is based on their very good knowledge of dynamic soil properties (mainly Vs values), and the knowledge of the depth of the bedrock basement. In this paper we present a selec ted sample of about 300 records. They were provided from Japan (164 KiKNet records), Greece, (58 records) and from USA California (86 strong ground re cords from the COSMOS Virtual Data Center and the National Strong Motion Project Datasets) (Fig. 2a). The seismic records are classified in two main categor ies depending on the level of seismic intensity, Mw>5.5 (or approximately PGA>0.2g) and Mw0.2) and Mw5.5 Normalized Acceleration Spectral Values
Soil Calss B M>5.5 62.75.5
MEAN VALUES
Soil Class B M5.5, n=92 records and (b) for M5.5), the design demand spectra of EC8 and the proposed demand spectra by Pitilakis et al. (2004) are better correlated with the mean values +1SD (Fig. 4). For earthquakes with M5.5 there are 70 records; 28 records from Japan, 14 from the Greece and 28 from California. The range of PGA values varies between 59gal and 741gal. For M5.5 595.5, n=70 records): (a) Absolute and (b) Normalized demand spectra Soil Class C M>5.5
MEAN VALUES 4
EC8 C (CEN 2004), TYPE 1 C1 TYPE 1
3
C2 TYPE 1
2
1
Normalized Acceleration Spectral Values
MEAN VALUES + 1SD
Normalized Acceleration Spectral Values
4
AVER. DEMAND SP.
Soil Class C M5.5, n=92 records and (b) for M5.5. The peak displacement spectral values are between 2.020.0cm, while the range of the maximum acceleration spectral values is between 5001500 cm/sec2. Equally large scatter of spectral values is also observed at the normalized curves of the demand spectra (Fig. 6). From Fig. 5 it is observed that the mean values of the computed demand spec tra for M>5.5 are well compared with the corresponding design demand spectra of EC8 and the proposed demand spectra by Pitilakis et al. (2004), especially for the spectral values which correspond to the constant plateau. For small earthquakes with M
Lihat lebih banyak...
Comentarios