Professional Documents
Culture Documents
www.elsevier.com/locate/engstruct
Received 20 September 1999; received in revised form 19 June 2000; accepted 20 June 2000
Abstract
This study describes numerically, the interdependency between several seismic acceleration parameters and diverse structural
damage indices. Peak ground motion, spectral and energy parameters are used for characterising the seismic excitation. On the
other hand both, structural and nonstructural damage is considered, expressed by the modified Park/Ang overall structural damage
index (OSDI), the maximum interstory drift (ISD) and the maximum floor acceleration. After the numerical evaluation of several
seismic parameters, nonlinear dynamic analyses are conducted to furnish the structural damage status. The degree of the interrelation-
ship between the seismic parameters and the damage indices is provided by correlation coefficients. The investigation is carried
out for a reinforced concrete plane frame system designed after Eurocodes 2 and 8 (EC2, EC8) and the aim is to determine the
characteristics of the accelerograms that exhibit the strongest influence on structural and nonstructural damages. The numerical
results have shown, that peak ground motion seismic parameters provide poor or fair correlation with the OSDI, whereas the spectral
and energy parameters provide good correlation. Furthermore, the central period and the strong motion duration after Trifunac/Brady
exhibit poor correlation with the OSDI. All these results give reason to recommend the spectra and energy related seismic intensity
parameters as reliable descriptors of the seismic damage potential. 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Ground motion; Acceleration parameter; Damage potential; Damage indices; Reinforced concrete
0141-0296/01/$ - see front matter 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 1 4 1 - 0 2 9 6 ( 0 0 ) 0 0 0 7 4 - 2
A. Elenas, K. Meskouris / Engineering Structures 23 (2001) 698–704 699
parameters which can be used to characterise the differ- Table 2 shows the seismic parameters for all seismic
ent forms of damage usually encountered. These damage excitations which are described in Table 1. These para-
indicators are presented in Section 2. Finally, correlation meters are: PGA (amax), PGV (vmax), PGD (dmax), SA,
coefficients are evaluated to express the grade of interde- SV, SD, CP T0, seismic input energy Einp, ARIAS inten-
pendency between seismic acceleration parameters and sity, SMD T0.90 and power P0.90. The spectral values SA,
several damage indices. The correlation coefficients SV, SD and Einp are given for the period of 1.18 s, which
between the different damage indices have also been is identical with the first eigenperiod of the investigated
evaluated. The presented methodology is applied to an frame structure, as presented in the Section 3.
eight storey reinforced concrete frame building subjected
to several well-known acceleration records.
3. Dynamic analysis of a frame structure
2. Seismic acceleration parameters The reinforced concrete frame structure shown in Fig.
1 has been designed according to the rules of the current
In the first step of the method just mentioned, the fol- Eurocodes for concrete structures and aseismic design
lowing seismic parameters are evaluated: peak ground (EC2 and EC8 respectively). The cross sections of the
acceleration, velocity and displacement (PGA, PGV, beams are T-shapes with 40 cm width, 20 cm plate thick-
PGD), spectral pseudo-acceleration (SA), spectral ness, 60 cm total beam height and 1.45 m effective plate
pseudo-velocity (SV), spectral displacement (SD), cen- width. The distance between frames in the three dimen-
tral period (CP), seismic input energy Einp, ARIAS inten- sional structure has been chosen to be 6 m. The building
sity, SMD, power P0.90 and the HUSID diagram. The has been considered as an ‘importance class III, ductility
definition of each parameter is presented in the literature class L’-structure according to Eurocode 8, with subsoil
[1–6,11] and will not be repeated here. Table 1 shows of type B and regional seismicity of category 3.
the data of the acceleration time histories which have Additionally to the dead weight and the seismic loading,
been selected for use in the present analysis. The events snow, wind and live loads have been taken into account,
have been chosen from worldwide well known sites with as well as lateral loads due to column sway. The first
strong seismic activity. The primary aspect which has natural period of the frame was 1.18 s. The ordinate of
been taken into consideration was to relate structural the used elastic response spectrum after EC8 was 0.20
damage to the damage potential of seismic excitations g. After designing and detailing the reinforced concrete
recorded at different soil conditions and source dis- frame structure, a nonlinear dynamic analysis was car-
tances. The seismic parameters of the accelerograms ried out for the evaluation of the structural seismic
have been evaluated by computer [4,11]. response. For this purpose the computer program
Table 1
Input ground motions
Table 2
Strong motion parameters
Earthquake PGA PGV PGD amax/vmax CP T0 ARIAS SMD Power Einp SA (g) SV SD (cm)
amax (g) vmax dmax (m) (g/m/s) (s) Intensity T0.90 (s) P0.90 (m2/s2) (cm/s)
(m/s) [m2/s3] (m2/s4)
Alkion (L) 0.240 0.225 0.099 1.067 0.422 4.334 15.660 0.249 0.001 0.034 0.550 0.100
Alkion (T) 0.296 0.253 0.071 1.170 0.381 5.326 13.990 0.343 0.001 0.218 37.550 6.642
Big Bear (270°) 0.472 0.282 0.046 1.674 0.130 19.963 9.720 1.848 0.055 0.096 15.710 2.926
Big Bear (360°) 0.534 0.344 0.043 1.552 0.128 17.154 10.360 1.490 0.824 0.172 29.727 5.278
Erzincan (N–S) 0.399 0.976 1.360 0.409 0.479 9.729 10.450 0.839 2.280 0.799 147.004 27.608
Erzincan (E–W) 0.501 0.829 0.908 0.605 0.339 11.381 7.470 1.372 0.959 0.527 96.480 18.089
Izmir (N–S) 0.417 0.123 0.084 3.401 0.194 2.168 0.830 2.364 0.004 0.019 3.441 0.644
Izmir (E–W) 0.139 0.068 0.031 2.050 0.158 0.406 1.590 0.229 0.002 0.017 3.152 0.589
Hyogo-Ken Nanbu 0.837 0.924 0.469 0.905 0.302 50.735 8.320 5.556 3.713 0.987 179.040 33.310
Kalamata 0.273 0.231 0.056 1.183 0.322 4.585 6.370 0.652 0.275 0.231 42.470 8.002
Montenegro 0.459 0.432 0.758 1.063 0.267 27.864 10.600 2.375 0.973 0.322 57.910 10.682
Landers (0°) 0.268 0.271 0.010 0.990 0.408 10.106 30.840 0.295 1.960 0.458 80.800 14.660
Landers (90°) 0.278 0.427 0.157 0.651 0.351 14.461 28.220 0.461 2.079 0.537 93.940 16.904
Cape Mendocino 1.468 1.261 0.360 1.164 0.177 36.242 13.540 2.448 1.320 0.630 106.475 19.215
(0°)
Cape Mendocino 1.019 0.405 0.150 2.517 0.221 14.481 13.920 0.937 0.315 0.292 50.562 8.997
(90°)
Naghloo 0.710 0.859 0.451 0.826 0.172 17.750 2.200 7.406 0.350 0.368 63.250 11.186
San Salvador (0°) 0.421 0.455 0.134 0.925 0.300 5.369 3.960 1.238 0.435 0.422 73.318 13.095
San Salvador (90°) 0.339 0.323 0.042 1.050 0.300 5.866 2.760 1.959 0.146 0.260 44.291 7.758
Strazhitsa 0.362 0.632 0.122 0.572 0.168 1.420 3.220 0.402 0.005 0.034 5.753 1.010
Whittier 0.607 0.277 0.025 2.191 0.137 7.881 2.665 2.661 0.076 0.125 21.548 3.826
冪冘 冘
N N
acceleration for all excitations. The degree of damage
can be classified according to the criteria presented in (Xi−X̄) 2
(Yi−Ȳ) 2
i⫽1 i⫽1
Table 4 [18]. An OSDI equal zero denotes that the struc-
ture remains in the elastic region during the excitation. where: X̄ and Ȳ are the mean values of Xi and Yi data
As mentioned in the previous section, the ordinate of respectively and N is the number of pairs of values
the used elastic response spectrum after EC8 was 0.20 (Xi,Yi) in the data.
g for the design of the frame. Taking this value as a The Spearman rank correlation coefficient between
threshold and comparing the column with the spectral two variables X and Y, is given by the relation:
702 A. Elenas, K. Meskouris / Engineering Structures 23 (2001) 698–704
Table 4
Damage classification limits
Damage
冘
N Table 6
6 D2 Rank correlation coefficients after Spearman
rSpearman⫽1⫺
i⫽1
(2) Seismic parameter OSDI Max. ISD Max. floor
N(N −1)
2
acceleration
where D denotes the differences between the ranks of
PGA 0.482 0.515 0.695
corresponding values of Xi and Yi and N is the number PGV 0.752 0.801 0.749
of pairs of values (X,Y) in the data. PGD 0.623 0.645 0.531
Table 5 presents the correlation coefficients after Pear- SA 0.869 0.836 0.665
son between all seismic parameters presented in Table 2 CP 0.319 0.265 0.005
and the damage indices. Through the Pearson correlation Einp 0.897 0.810 0.700
ARIAS intensity 0.727 0.698 0.785
coefficient it can be seen that the OSDI and the max. SMD 0.331 0.188 0.099
ISD have the same grade of interrelation to the seismic Power P0.90 0.461 0.568 0.738
parameters. Both of them have the maximum correlation amax/vmax ⫺0.511 ⫺0.551 ⫺0.305
to SA and Einp, while minimum correlation occurs with
respect to SMD and CP. The maximum floor acceler-
ation exhibits best correlation with the ARIAS intensity,
lation to SA and Einp, while minimum correlation occurs
while minimum correlation is present, as in the case of
in respect to SMD and CP. The maximum floor acceler-
the OSDI and ISD before, with SMD and CP. It must
ation shows maximum correlation to the ARIAS inten-
be pointed out that the SMD as defined by Trifunac and
sity, PGV and power P0.90. Its minimum correlation
Brady shows very poor correlation with the damage indi-
occurs, as for OSDI and ISD before, for the SMD and
ces. This is due to the fact that its definition does not
CP values.
take into account the absolute seismic energy value,
Table 7 shows the correlation coefficients after Pear-
being based only on its temporal distribution.
son and the rank correlation coefficient after Spearman
By examining Spearman’s rank correlation coefficient
for the examined damage indices. Observing the corre-
in Table 6, we recognise that the OSDI and the max.
lation coefficients after Pearson, we recognise the strong
ISD have the same grade of interrelation to the seismic
correlation between OSDI and ISD and the fair corre-
parameters as in the case of the Pearson correlation coef-
lation between maximum floor acceleration and OSDI as
ficient. These two parameters exhibit maximum corre-
well as ISD. The rank correlation is marked for all three
damage indices.
Table 5 The quantities, which furnish a coefficient greater than
Correlation coefficients after Pearson 0.561 in Tables 5–7 exhibit a significant correlation at
Seismic parameter OSDI Max. ISD Max. floor
acceleration Table 7
Correlation coefficients between the damage indices
PGA 0.383 0.419 0.717
PGV 0.679 0.651 0.728 Pearson correlation OSDI 1.000
PGD 0.648 0.531 0.466 Max. ISD 0.970 1.000
SA 0.910 0.854 0.692 Max. acc. 0.695 0.652 1.000
CP 0.373 0.253 0.057
Einp 0.878 0.839 0.504 ODSI ISD Max. acc.
ARIAS intensity 0.753 0.802 0.752
SMD 0.154 0.038 0.119 Spearman correlation Max. acc. 0.870 0.943 1.000
Power P0.90 0.445 0.512 0.508 Max. ISD 0.944 1.000
amax/vmax ⫺0.432 ⫺0.354 ⫺0.341 OSDI 1.000
A. Elenas, K. Meskouris / Engineering Structures 23 (2001) 698–704 703
6. Conclusions
period and the strong motion duration after parameters. In: Proceedings of the 10th European Conference on
Trifunac/Brady exhibited poor correlation with the Earthquake Engineering, Vienna 1994. Rotterdam: Balkema,
1995:1011–6.
OSDI. Due to this, spectral and energy related para- [8] Elenas A, Liolios A. Earthquake induced nonlinear behavior of
meters are believed to be better suited for the characteris- reinforced concrete frame structures in relation with characteristic
ation of the seismic damage potential. The HUSID dia- acceleration parameters. In: Proceedings of the 5th International
gram, although being a structure-independent seismic Conference on Seismic Zonation, Nice 1995. Nantes: Presses
parameter, provides a good estimation of the overall Académiques, 1995:1013–20.
[9] Elenas A. Interdependency between seismic acceleration para-
structural damage evolution time history. The Pearson meters and the behaviour of structures. Soil Dynam Earthq Engng
correlation is strong between OSDI and max. ISD and 1997;16(5):317–22.
medium between max. acceleration and OSDI as well as [10] Elenas A. Interrelation between seismic acceleration parameters
max. ISD. On the other hand, the rank correlation after of European earthquakes and the structural behavior. Eur Earthq
Spearman is strong for all three damage indices. While Engng 1998;12(1):56–63.
[11] Mahin SA, Lin J. Construction of inelastic response spectra for
these tendencies in the correlation study are deemed to single-degree-of-freedom systems. Report no. UCB/EERC-83/17.
be quite well-established, further investigations utilising University of California at Berkeley: Earthquake Engineering
a larger accelerogram database would lead to greater Research Center, 1983.
statistical validity and also enhance the quantitative [12] Reinhorn AM, Kunnath SK, Valles-Mattox R. IDARC 2D ver-
aspects. sion 4.0: users manual. State University of New York at Buffalo:
Department of Civil Engineering, 1996.
[13] Priestley MJN. Displacement-based approaches to rational limit
states design of new structures. In: Proceedings of the 11th Euro-
pean Conference on Earthquake Engineering. Volume of invited
References lectures, Paris 1998. Rotterdam: Balkema, 1998:317–35.
[14] Rodriguez-Gómez S, Çakmak AS. Evaluation of seismic damage
[1] Jennings PC. Engineering seismology. In: Proceedings of the indices for reinforced concrete structures. Technical report
International School of Physics ‘Enrico Fermi’, Course LXXXV, NCEER-90-0022. State University of New York at Buffalo:
Earthquakes: Observation, Theory and Interpretation, Varenna Department of Civil Engineering, 1990.
1982. Bologna: Italian Physical Society, 1982:138–73. [15] DiPasquale E, Çakmak AS. On the relation between local and
[2] Arias A. A measure of earthquake intensity. In: Seismic design global damage indices. Technical report NCEER-89-0034. State
for nuclear power plants. Cambridge, MA: MIT Press, University of New York at Buffalo: Department of Civil Engin-
1970:438–69. eering, 1989.
[3] Trifunac MD, Brady AG. A study on the duration of strong earth- [16] Garstka B, Krätzig WB, Meskouris K, Meyer IF, Stangenberg
quake ground motion. Bull Seismol Soc Am 1975;65(3):581–626. F. Damage assessment in cyclically loaded reinforced concrete
[4] Meskouris K. Structural dynamics. Berlin: Ernst and Sohn, 2000. columns. In: Proceedings of the European Conference on Struc-
[5] Naeim F, Anderson JC. Classification and evaluation of earth- tural Dynamics EURODYN’90, Bochum 1990. Rotterdam: Balk-
quake records for design. EERI/FEMA NEHRP Fellowship ema, 1991:551–6.
report PF-2/93. Berkeley: EERI, 1993. [17] Park YJ, Ang AH-S. Mechanistic seismic damage model for
[6] Trifunac MD, Novikova EI. State of the art review on strong reinforced concrete. J Struct Engng 1985;111(4):722–39.
motion duration. In: Proceedings of the 10th European Confer- [18] Gunturi SKV, Shah HC. Building specific damage estimation.
ence on Earthquake Engineering, Vienna 1994. Rotterdam: Balk- In: Tenth World Conference on Earthquake Engineering, Madrid
ema, 1995:131–40. 1992. Rotterdam: Balkema, 1992:6001–6.
[7] Elenas A, Liolios A, Vasiliadis L. Earthquake induced nonlinear [19] Spiegel MR. Theory and problems of statistics. London:
behavior of structures in relation with characteristic acceleration McGraw-Hill, 1992.