Comparison of real and simulated records using ground motion intensity measures

https://doi.org/10.1016/j.soildyn.2021.106796Get rights and content

Highlights

  • Differences between seismic intensity measures for simulated and real records are investigated.

  • Alternative intensity parameters are evaluated for both real and simulated record sets.

  • Varying patterns are observed for different levels of ground motion intensity parameters.

  • Correlations of intensity measures with seismic response of SDOF models are investigated.

  • Better correlations are observed for simulated data emphasizing their significance.

Abstract

Simulated ground motion records have recently become major alternatives to real records. In this study, differences between simulated and real record sets are evaluated in terms of alternative ground motion intensity measures. As case studies, two regions in Turkey (Erzincan and Duzce) are considered. A large set of simulated records involving a wide range of magnitudes, source-to-site distances, and site properties is prepared. The fundamental seismological properties of stochastically-simulated records are compared with those of actual motions. A real ground motion dataset for each region with consistent source, distance, and site properties is formed. Alternative intensity parameters are computed for real and simulated sets and evaluated through statistical analyses involving Kolmogorov–Smirnov test. Varying patterns are observed for different levels of intensity parameters. Correlations of intensity measures with maximum inelastic displacements obtained from single-degree-of-freedom models are investigated. Better correlations are observed for simulated data emphasizing significance of region-specific ground motions.

Introduction

Estimation of fatalities and economic losses during potential seismic activities is directly related to reliable seismic performance assessment of structures. Accordingly, proper selection and utilization of ground motion records for seismic performance evaluation of structures has gained much attention during the last decade [[1], [2], [3], [4]]. Ground Motion Intensity (GMI) measures have been employed in order to practically reflect the effects of full time series of ground motion records [[5], [6], [7], [8], [9], [10]]. These measures are employed for various purposes such as performance-based seismic assessment, damage and loss estimation as well as characterizing seismic response of special structures [7,[11], [12], [13]]. The GMIs are computed for either regional acceleration data or records of other regions with similar seismicity. During the selection of ground motion record sets for vulnerability analysis, a range of GMI measures is considered. The criteria to select the ground motions requires a nearly uniform distribution of ground motions in terms of the GMI in the dataset. Since a uniform distribution is not generally possible, scaling is applied to fill the GMI gap in the selected ground motion set [14]. Therefore, not only the choice of GMI but also its ability to represent the characteristic of the ground motion records is critical.

An alternative and efficient way is to use simulated ground motions compatible with seismological characteristics of the area to form regional ground motion sets with a uniform distribution of GMI. Such an effort would involve alternative methods of simulation with different levels of accuracy and computational time [[15], [16], [17], [18], [19], [20], [21], [22]]. Until now, seismological characteristics of the simulated records are investigated against the corresponding real data at existing stations that recorded the past events [[23], [24], [25], [26]]. Then, estimation of alternative engineering demand parameters (EDPs) based on the simulated records of past events is evaluated against the real demands of a particular earthquake [[27], [28], [29], [30], [31], [32], [33], [34], [35]]. However, a thorough comparison of seismic intensity measures from large real and simulated datasets is required in order to study the significance of regional datasets.

The main objectives of this study are twofold: 1) to investigate the consistency of the intensity measures computed from the real and (already validated) simulated ground motion records; 2) to study any potential correlations between these intensity measures and a selected engineering demand parameter. For this purpose, real and simulated sets are selected to be consistent with the hazard levels and seismotectonics of the areas studied. The real records are selected from PEER database [36] considering moment magnitude, source type, site characteristics, and source-to-site distances and the ranges of the GMIs employed. In order to ensure a nearly uniform GMI distribution in the selected ground motion data set, particularly for the larger values of GMIs, scaling is applied to the ground motions. Alternative to real records, ground motion simulations provide a platform for assessing regional tectonic and seismological characteristics. Therefore, simulated records for two different regions in Turkey (Erzincan and Duzce) along the North Anatolian Fault Zone (NAFZ) are selected from the synthetic database presented in Askan and Karimzadeh [23]. Simulated ground motion records are selected based on an almost uniform distribution of considered GMIs. Since detailed velocity models of soft soil media on NAFZ are not available which makes the use of deterministic approaches impractical to reach the frequencies of interest, among various simulation techniques, stochastic finite-fault methodology is employed. Despite the simplifications involved, this methodology effectively simulates the frequencies which affect the built environment [37,38].

The ground intensity measures (GMI) employed in this study are selected as the most commonly used ones among the existing intensity measures: Peak Ground Acceleration (PGA), Peak Ground Velocity (PGV), a modified version of Acceleration Spectrum Intensity (ASI*) [7], Housner Intensity (HI), Velocity Spectrum Intensity (VSI), Cumulative Absolute Velocity (CAV), Arias Intensity (Ia) and Significant Duration (td). These GMIs are computed for the selected real and simulated records for each region. The GMIs from real and simulated ground motion records are also compared through Pearson correlation coefficient. Finally, a wide range of Single-Degree-Of-Freedom (SDOF) analyses are carried out to determine the maximum inelastic displacements due to all of the ground motion sets and the correlation of GMI with maximum inelastic displacement is investigated.

Section snippets

Regional ground motion simulations in the study areas

The simulated ground motions used in this study are obtained with the stochastic finite-fault method based on a corner frequency approach within the EXSIM platform [37]. In the stochastic finite-fault method, the rectangular fault plane is subdivided into sub-faults where each sub-fault is modeled as a stochastic point-source [17,39]. Then, the sub-source contributions are summed up in time domain by kinematic delays to model the final acceleration response. The ground motion acceleration

Selection of real and simulated ground motion records

Alternative simulated and real databases are formed for each region according to different GMIs. While constructing these databases, regional seismological characteristics in terms of magnitude, source-to-site distances and soil types are considered.

COMPARISON of ground motion sets

In this part, the real and simulated ground motion sets for both regions are evaluated in terms of the GMI parameters employed. The intensity measures are calculated for both real and simulated records in terms of PGA, PGV, ASI, ASI*, HI, VSI, CAV, Ia and td. Fig. 2, Fig. 3 present the scatter diagrams of each GMI, corresponding to the PGA-based (Set 1E vs. 3E) and PGV-based (Set 2E vs. 4E) record sets for Erzincan region, respectively. Similarly, Fig. 4, Fig. 5 illustrate the scatter diagrams

Conclusions

In this study, simulated ground motion record sets are compared with the corresponding real record sets in terms of commonly used ground motion intensity measures in order to evaluate the efficiency of region-specific simulations. For this purpose, the real record sets for each study area are selected such that the sets are seismologically compatible with the regional seismicity of the corresponding region. Eventually, while forming the datasets, consistent regional source properties,

Author statement

Shaghayegh Karimzadeh: Ground motion simulation, Seismic Intensity evaluation, Data Analysis, Writing- Original draft preparation.

Koray Kadas: Correlation studies, Real ground motion selection.

Aysegul Askan: Supervision, Writing, Reviewing and Editing.

Ahmet Yakut: Supervision, Writing, Reviewing and Editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The ground motion dataset in Erzincan was prepared within the Turkish National Union of Geodesy and Geophysics project with grant number TUJJB-UDP-01-12. The second author has been financially supported by The Scientific and Technological Research Council of Turkey (TUBITAK) through Domestic Doctoral Scholarship Program.

References (49)

  • P. Giovenale et al.

    Comparing the adequacy of alternative ground motion intensity measures for the estimation of structural responses

    Earthq Eng Struct Dynam

    (2004)
  • J.W. Baker et al.

    A vector‐valued ground motion intensity measure consisting of spectral acceleration and epsilon

    Earthq Eng Struct Dynam

    (2005)
  • A. Yakut et al.

    Correlation of deformation demands with ground motion intensity

    J Struct Eng

    (2008)
  • P. Tothong et al.

    Structural performance assessment under near‐source pulse‐like ground motions using advanced ground motion intensity measures

    Earthq Eng Struct Dynam

    (2008)
  • C. Huang et al.

    Ground‐motion intensity measure correlations observed in Italian strong‐motion records

    Earthq Eng Struct Dynam

    (2019)
  • Y.M. Wu et al.

    Relationships between strong ground motion peak values and seismic loss during the 1999 Chi-Chi, Taiwan earthquake

    Nat Hazards

    (2004)
  • Z. Karimi et al.

    Ground motion intensity measures to evaluate II: the performance of shallow-founded structures on liquefiable ground

    Earthq Spectra

    (2017)
  • S.H. Hartzell

    Earthquake aftershocks as Green's functions

    Geophys Res Lett

    (1978)
  • K. Irikura

    Semi-empirical estimation of strong ground motions during large earthquake

    (1983)
  • D.M. Boore

    Stochastic simulation of high-frequency ground motions based on seismological models of the radiated spectra

    Bull Seismol Soc Am

    (1983)
  • I.A. Beresnev et al.

    FINSI--a FORTRAN program for simulating stochastic acceleration time histories from finite faults

    Seismol Res Lett

    (1998)
  • H. Bao et al.

    Large-scale simulation of elastic wave propagation in heterogeneous media on parallel computers

    Comput Methods Appl Math

    (1996)
  • V. Akcelik et al.

    High-resolution forward and inverse earthquake modeling on terascale computers

  • R.W. Graves et al.

    Broadband time history simulation using a hybrid approach. Proceedings of 13th World Conference in Earthquake Engineering

    (2004)
  • Cited by (10)

    • Investigation of the effect of real ground motion record number on seismic response of regular and vertically irregular RC frames

      2022, Structures
      Citation Excerpt :

      On the other hand, ground motion intensity measures (IMs) are important parameters for estimating the structural response levels. In the literature, several studies can be found discussing correlation between IMs and seismic demands or seismic damage indices [14–17]. These studies are concentrated on the determination of the most correlated IMs with structural responses.

    View all citing articles on Scopus
    View full text