Académique Documents
Professionnel Documents
Culture Documents
Pavel]
On: 16 January 2014, At: 01:59
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
To cite this article: F. Pavel & D. Lungu (2013) Correlations Between Frequency Content Indicators
of Strong Ground Motions and PGV, Journal of Earthquake Engineering, 17:4, 543-559, DOI:
10.1080/13632469.2012.762957
To link to this article: http://dx.doi.org/10.1080/13632469.2012.762957
1. Introduction
Ground motions records show various frequency content, from wide and/or intermediate
frequency bandwidth ground motions (in hard and/or medium soil conditions) to narrow
frequency band ground motions (in soft soil conditions). The random frequency content
of the recorded strong ground motions generally depends on both magnitude and source
mechanism as well as on local soil conditions and epicentral distance.
Probability-based assessment [Lungu and Cornea, 1987; Lungu et al., 1993] of the frequency content of the ground motion records can be done using the Power Spectral Density
(PSD) and its related indicators: the dimensionless bandwidth indicators (Cartwright
& Longuet-Higgins) and q (Vanmarcke), as well as fractile frequencies f10 , f50 , and f90
(KennedyShinozuka indicators) below which 10%, 50%, and 90% of the total cumulative
power of the PSD occurs.
The best deterministic indicators of the frequency content of the ground motion records
are the control periods of structural response spectra, historically introduced by Newmark
and Hall [1969, 1982]. The evolution of this concept during the years 19782010 is further
described in ATC [3-06, 1978]; Lungu et al. [1997]; Rathje et al. [1998]; Bommer et al.
[2000]; and ASCE [7-10, 2010].
The amplitude of the ground motion can be usually described by the peak ground
acceleration and/or by the peak ground velocity (PGV) which has received less attention
than PGA. Due to the fact that PGV is less sensitive to the high frequency amplitudes
Received 19 June 2012; accepted 23 December 2012.
Address correspondence to F. Pavel, Department of Reinforced Concrete Structures, Technical University
of Civil Engineering Bucharest, Bd. Lacul Tei, 122-124, RO 020396, Romania. E-mail: florin.pavel@utcb.ro
543
544
of the ground motion, it can be used to characterize the damage potential of the seismic
record. PGV has been linked [Akkar and Ozen, 2005] to the earthquake magnitude, strong
ground motion duration, frequency content of ground motions, and inelastic deformation
demands in SDOF (single degree of freedom) systems. Based on their results the authors
consider PGV as a stable candidate for ground motion intensity measure in simplified
seismic assessment methods. Yakut and Yilmaz [2007] presented a series of analyses on
MDOF (multiple degree of freedom) systems from which it appears that PGA is a better
GMI (ground motion intensity) indicator than PGV. However, the conclusion of the authors
is Since PGA and PGV do not reflect the characteristics of the structure they appear to be
inadequate intensity parameters. Fajfar et al. [1990] introduced a damage potential indicator for intermediate period structures based on PGV and strong ground motion duration,
while Bommer and Alarcn [2006] and Booth [2007] introduced relations for estimating
PGV using random vibration theory. The relation between the Modified Mercalli Intensity
(MMI) and the peak ground acceleration PGA and peak ground velocity was studied in
Wald et al. [1999].The ratio PGA/PGV can be used both as a frequency content indicator
of the strong ground motion and as a damage indictor as outlined in Zhu et al. [1988].
Elnashai and Di Sarno [2008] divided the ratio into 3 categories from low ratio (PGA/PGV
< 0.8) to high ratio (PGA/PGV > 1.2). The lower bound for potentially damaging PGV
was established in Bommer and Martinez-Pereira [2000] at a value of 20 cm/s.
In the present article, a comparative analysis of stochastic and deterministic frequency
content indicators is applied to a set of 30 strong ground motion records having PGA from
0.20.9 g. Also, the relation between PGV and the spectral acceleration (SA) of the selected
ground motion is analyzed. Finally, the relation between PGV and the ratio PGA/PGV on
one hand and a series of frequency content indicators is considered.
[a(t)]2 dt.
(1)
Consequently, the power spectral density (PSD) of accelerograms considered in the present
study was determined for the stationary part of the record modeled to be within the time
interval t0.10 t0.90 .
The dimensionless indicator proposed by Cartwright & Longuet-Higgins and defined
in Clough and Penzien [2003] as a function of the spectral moments of the PSD for the
stationary process of the ground acceleration:
0=
22
1,
0 4
(2)
545
i = i Sx () d.
(3)
The guidance values for indicator in the case of actual ground motion accelerograms
might be [Clough and Penzien, 2003]:
2/3 < < 0.85 for a wide frequency band process (white noise or band-limited
white noise);
0.85 < < 0.90 for an intermediary band process;
> 0.90 for narrow frequency band processes (very high-frequency low-amplitude
signal associated with a low-frequency band-limited signal).
Several representative examples of ground motions from Romania are: the narrowest frequency band ground motions ever recorded in Romania at INCERC station in Bucharest
(the NS comp. of the 1977 and 1986 seismic events) and the broadest frequency band
ground motions recorded at Carcaliu in Dobrogea during the 1986 and 1990 Vrancea earthquakes. Another interesting series of seismic motions was recorded at Cernavoda City Hall
during the same 1986 and 1990 events; these represent ground motions with a very stable
predominant period.
In the epicentral area of the Vrancea earthquakes and in the East of the Carpathian
Mountains (Moldova), the recorded ground motions are characterized by wide frequency
bandwidth: the maximum ordinates of the response spectra are in the range T = 0 0.6 s.
In the Bucharest region, the recorded ground motions can be divided into two categories
[Lungu et al., 1992]:
seismic records characterized by spectral peaks in the long period range (T > 1.0 s)
and specific to narrow frequency bandwidth processes; or
seismic records having the maximum spectral peaks in the short period range (T <
0.6 s), characteristic to wide frequency bandwidth processes.
546
EPV
EPA
SV 0.81.2s
2.5
EPD
EPV
EPV
EPA
EPD =
EPV =
V PGV
=
A PGA
TD = 8
TC = 5
PGD
PGV
PGV
PGA
Bommer et al.
[2000]
PGD
1.39 PGD
= 5.29
= 2
1.65 PGV
PGV
V PGD
=
A PGV
PGV
1.65 PGV
= 4.89
2.12 PGA
PGA
TD = 2
= 2
TC = 2
(1) Definitions based on a fixed period window for computing EPA (0.10.5 s) and EPV (0.81.2 s).
(2) Definitions based on a mobile period window (of 0.4 s width) for getting maximum effective values.
max SV 0.4
2.5
max SD0.4
2.5
(2)
SA0.4
EPA = max2.5
TD = 2
TC = 2
EPV =
SA0.10.5s
EPA =
2.5
(1)
TC = 2
SD1
SDS
TC =
547
3.0
Amplification factor
2.5
2.0
1.5
1.0
0.5
0.0
TB = 0.1 TC
TC
TD
Period, s
FIGURE 1 Code spectral shape from the Romanian seismic code P100-1/2006 [2006].
The definitions in ATC 3-06 [1978] and Lungu et al. [1997] are based on the spectral values
and use the effective peak acceleration, velocity, and displacement which represent averaged values, while the American Code [ASCE 7-10, 2010] uses the spectral acceleration
values at two periods corresponding to the short period range (0.2 s) and to the medium
period range (1.0 s).
Figure 1 shows an example of a code spectrum from the Romanian seismic code P1001/2006 [2006]. The control period T B is taken equal to 0.1T C .
The mean period T M is computed based on the relations given in Rathje et al. [1998]:
2
C /fi
TM = i 2 ,
Ci
(5)
where Ci represent the Fourier amplitudes of the entire accelerogram and fi are the discrete
Fourier transform frequencies between 0.25 20 Hz.
548
8.5
Magnitude,Mw
8
7.5
Kobe 1995
7
6.5
6
Kalamata 1986
5.5
Ancona 1972
5
4.5
150
300
450
600
750
900
120
Iwate 2008
Vrancea
Kushiro 1993
1977
India-Myanmar 1988
Depth, km
100
80
60
Miyagi 2003
Maule 2010
40
Manjil 1990
Kobe 1995
Bam 2003
Van-Merkez 2011
20
0
150
300
450
PGA,
600
900
750
cm/s2
8
7
Number of events
PGA, cm/s2
6
5
4
3
2
1
0
100
200
300
400
500
PGA,
600
700
cm/s2
800
900
549
mean = 43 cm/s
COV = 0.56
12
Number of events
10
8
6
4
2
20
40
60
PGV, cm/s
80
100
120
Number of events
12
10
8
6
4
2
0
4
Magnitude, MW
14
12
Number of events
10
8
6
4
2
0
0
10
20
30
40
Depth, km
40 130
550
12
Number of events
10
8
6
4
2
0
0
20
40
60
80
Epicentral distance, km
6
5
4
3
2
1
0
1940 1970
1970
1980
1990
2000
2010
2011
Year
4. PGV Estimation
Another focus of this study is the computation of the ratio PGV/SA for the analyzed ground
motions, as presented in Bommer and Alarcn [2006]. In Booth [2007] are given different
relations for computing PGV based on spectral values and on random vibration theory.
The ratio PGV/SA for the 30 seismic records used in this article and also the mean
ratio PGV/SA are shown in Fig. 10. The values of the coefficient of variation of the ratio
PGV/SA are shown in Fig. 11. The ratio proposed by Bommer and Alarcn [2006] is the
following:
PGV(cm/s) =
SA0.5s (cm/s2 )
.
20
(6)
It is clear from the above 2 figures that the minimum value of the ratio PGV/SA is encountered at a period of around 1.0 s. The minimum value of the coefficient of variation is 0.40.
The correlation between the ratio SA1.0 /8 and PGV is depicted in Fig. 12.
551
Miyagi, Morioka NS
Ancona, Ancona Rocca NS
Iwate, Hachinohe EW
Manjil, Abbar Transv.
Miyagi, Tohno NS
Tarapaca, Cuya Transv.
Umbria Marche, Nocera Umbra NS
Duzce, Duzce IRIGM NS
India-Myanmar, Diphu NS
Van Ercis, Muradiye NS
Bam, Bam N08E
Friuli, Tolmezzo NS
Kushiro, Kushiro JMA N153E
Aegion, Aegion Long.
Imperial Valley, El Centro NS
Kocaeli, Duzce EW
Vrancea, Petresti Focsani NS
Earthquake record
0.66
0.70
0.72
0.74
0.76
0.78
0.79
0.81
0.81
0.82
0.82
0.83
0.85
0.87
0.88
0.88
0.89
8.16
5.99
4.27
6.67
5.49
4.58
5.74
3.47
3.88
2.29
4.55
3.34
3.73
2.29
2.64
1.65
2.18
0.15
0.19
0.16
0.35
0.16
0.17
0.34
0.31
0.22
0.56
0.46
0.39
0.44
0.50
0.65
0.75
0.66
0.16
0.09
0.20
0.52
0.27
0.21
0.32
0.44
0.30
0.74
0.47
0.32
0.32
0.39
0.47
0.75
0.65
0.23
0.20
0.34
0.64
0.38
0.31
0.33
0.40
0.43
0.63
0.75
0.39
0.46
0.53
0.65
0.97
0.84
Lungu
et al.
[1997]
0.16
0.09
0.21
0.53
0.28
0.21
0.33
0.45
0.31
0.75
0.48
0.33
0.33
0.40
0.48
0.76
0.67
Boomer
et al.
[2000]
Control period TC , s
Freq.
Newmark
content Mean freq.
& Hall
indicator, f50 , Hz
ATC 3-06
[1982]
Stochastic indicators
Deterministic indicators
TABLE 2 Stochastic and deterministic indicators for analysis of the frequency content of selected ground motions
0.10
0.05
0.10
0.38
0.09
0.07
0.18
0.19
0.11
0.88
0.27
0.40
0.41
0.56
0.77
0.97
0.70
ASCE
7-10
[2010]
(Continued)
0.16
0.15
0.25
0.30
0.25
0.23
0.21
0.33
0.31
0.58
0.40
0.40
0.32
0.50
0.53
0.87
0.68
Mean period
TM , s
Rathje et al.
[1998]
552
Earthquake record
TABLE 2 (Continued)
0.90
0.90
0.91
0.93
0.93
0.94
0.95
0.95
0.96
0.97
0.98
0.98
0.99
1.92
1.45
1.76
1.53
1.61
1.36
1.32
1.64
0.74
0.79
0.58
0.54
0.46
0.40
0.86
0.86
0.92
0.99
0.93
0.97
0.49
1.04
1.02
1.21
1.14
0.81
0.54
0.47
0.49
0.49
0.97
0.72
0.68
0.83
1.30
0.74
1.67
1.02
1.76
0.49
0.74
0.75
0.70
1.41
0.85
0.74
1.04
1.51
1.28
1.49
1.55
2.21
Lungu
et al.
[1997]
0.55
0.48
0.50
0.50
1.00
0.73
0.69
0.85
1.33
0.75
1.70
1.04
1.80
Boomer
et al.
[2000]
Control period TC , s
Freq.
Newmark
content Mean freq.
& Hall
indicator, f50 , Hz
ATC 3-06
[1982]
Stochastic indicators
Deterministic indicators
0.43
0.53
1.48
0.77
1.28
1.06
1.22
0.82
1.25
1.07
1.88
1.44
1.28
ASCE
7-10
[2010]
0.56
0.61
0.64
0.64
0.84
0.74
0.66
0.93
1.32
1.17
1.43
1.54
2.13
Mean period
TM , s
Rathje et al.
[1998]
553
PGV/SA, s
0.8
0.6
0.4
0.2
0
0
0.5
1.5
Period, s
FIGURE 10 Ratio PGV/SA for the 30 analyzed ground motions (color figure available
online).
1.2
1
COV
0.8
0.6
0.4
0.2
0
0
0.5
1.5
Period, s
FIGURE 11 Coefficient of variation of the ratio PGV/SA for the 30 analyzed records.
120
r = 0.88
100
SA1.0/8, cm/s2
80
60
40
20
0
0
20
40
60
80
100
PGV, cm/s
120
554
The value of the correlation coefficient r is around 0.9, so on the basis of the investigation
of 30 ground motion records the approximate relation for the computation of PGV is:
PGV(cm/s) =
SA1.0s (cm/s2 )
.
8
(7)
Several correlations between the control periods TC determined according to the definitions
given by various authors are plotted in Figs. 13ad. The correlation coefficient r in Figs.
13ad may also suggest the definitions to be used in practical design. The large values of
the correlation coefficients prove the fact that the relations used for the computation of the
control period TC yield similar results.
Another correlation which has to be made is between the control period T C and the
mean period T M as shown in Figs. 14a and b.
The values of the correlation coefficients are similar proving the fact that the two
methods for TC and TM based on spectral values and, respectively, based on the Fourier
Amplitude Spectrum (FAS) provide similar results.
(b)
r = 0.93
1.5
1
0.5
0
0.5
1.5
2.5s
TC Lungu et al. (1997)
2s
1.5
1
0.5
0
2s
r = 0.94
0.5
1.5
2.5 s
2.5 s
r = 0.94
2
1.5
1
0.5
0
2.5 s
TC Lungu et al. (1997)
(c)
TC Lungu et al. (1997)
1.5
1
0.5
0
0.5
1.5
2.5 s
r = 0.81
0.5
1.5
2.5 s
(b)
2.5 s
555
r = 0.97
2
1.5
1
0.5
0
0
0.5
1.5
2.5 s
2.5 s
r = 0.93
2
1.5
1
0.5
0
0.5
TM
1.5
2.5 s
TM
r = 0.67
TC ASCE 7-10 (2010)
(b)
2.5 s
2
1.5
1
0.5
1.5
1
0.5
0
20
40
60
80
100
r = 0.76
0
0
120
20
40
60
80
100
120
PGV, cm/s
PGV, cm/s
(a)
2.5 s
2.5 s
r = 0.83
2
1.5
1
0.5
1.5
1
0.5
0
r = 0.94
PGA/PGV, gs/m
PGA/PGV, gs/m
(c)
2.5 s
(a)
r = 0.88
2
1.5
1
0.5
0
0
PGA/PGV, gs/m
FIGURE 16 (a)(c) Correlations between the ratio PGA/PGV and several T C definitions.
The above graphs prove the fact that the ratio PGA/PGV is more adequate parameter
for characterizing of the frequency content of ground motions than PGV. The correlation coefficients in Figs. 15 and 16 are in the range 0.670.77 for PGV and in the range
0.830.94 for PGA/PGV.
Figures 17a and b show the correlations between the mean period TM and PGV or the
ratio PGA/PGV.
The same conclusion is valid in the case of correlation of TM with PGV or PGA/PGV.
Moreover, the same conclusion applies in the case of the correlation of the stochastic
dimensionless indicator and PGV or PGA/PGV given in Figs. 18a and b.
Figures 19a and b present the correlation between the shear wave velocity VS,30 and the
mean period TM or the stochastic frequency bandwidths indicator . However, only 20 out
of 30 VS,30 values were obtained from various databases, such as: PEER strong motion
database, European strong motion database, Itaca strong motion database, or strong ground
motion database of Turkiye.
It is clear that there is a correlation between the values of the frequency content indicators and the value of the shear wave velocity VS,30 . However, the values of the correlation
coefficients are not very high and this may suggest that the parameter VS,30 may not adequately represent local site conditions. This is due to the fact that by using only the top
30 m of soil layers, other deeper soil layers are neglected. Furthermore, by constraining the
depth of considered soil layers at 30 m, a part of the layer situated around this depth may
be neglected (the part below 30 m).
(a)
(b)
2.5 s
2.5 s
r = 0.70
1.5
1.5
r = 0.83
TM
TM
0.5
0.5
0
0
20
40
60
80
100
120
PGA/PGV, gs/m
PGV, cm/s
(a)
1
0.9
0.8
0.7
0.6
r = 0.78
Dimensionless indicator,
Dimensionless indicator,
556
20
40
60
PGV, cm/s
80
100
120
r = 0.87
0.9
0.8
0.7
0.6
0
PGA/PGV, gs/m
FIGURE 18 (a)(b) Correlations between the dimensionless indicator and PGV or the
ratio PGA/PGV.
r = 0.61
TM
1.5
1
0.5
0
100
200
300
400
500
Vs,30, m/s
600
700
800
1
Dimensionless indicator,
(a)
557
r = 0.61
0.9
0.8
0.7
0.6
FIGURE 19 (a)(b) Correlations between the indicator Vs,30 and TM or the dimensionless
indicator .
In the case of Bucharest, in Calarasu [2012], no correlations where observed between
either VS,30 or VS,60 determined for 32 boreholes and the frequency content indicators
TC or .
7. Conclusions
The definitions for the control period of the response spectra TC given by Newmark
and Hall [1969, 1982], Lungu et al. [1997], and Bommer et al. [2000] are clearly
different, but very well correlated.
The mobile window procedure [Lungu et al., 1997] is regarded as the most adequate
instrument for computing the values of the control period TC of response spectra.
The values of the control period TC given by Newmark and Hall [1969, 1982], Lungu
et al. [1997], Bommer et al. [2000], and the values of the mean period TM given by
Rathje et al. [1998] are also well correlated. The value of the coefficient of correlation between TC given by Lungu et al. [1997] and the mean period TM given in
Rathje et al. [1998] is 0.97.
PGV can be approximately estimated from the spectral acceleration at 1.0 s as
(SA1.0 )/8; the ratio SA/PGV seems to be dependent on the frequency content of
the analyzed ground motions.
The ratio PGA/PGV is a better indicator of the frequency content of ground motion
than PGV.
The ratio PGA/PGV is very well correlated both with: (a) the deterministic frequency content indicators TC and TM , as well as with (b) the stochastic frequency
content indicator .
Acknowledgments
The authors would like to thank the reviewers for their very useful and constructive
opinions and suggestions regarding the article contents.
References
Akkar, S. and Ozen, O. [2005] Effect of peak ground velocity on deformation demands of SDOF
systems, Earthquake Engineering and Structural Dynamics 34, 15511571.
558
ASCE/SEI 7-10 [2010] Minimum design loads for buildings and other structures ASCE, American
Society of Civil Engineers, Reston, Virginia.
ATC 3-06, Special Publication 510/NSF [1978] The tentative provisions for the development of
seismic regulations for buildings, Prepared by ATC (Applied Technology Council) associated
with the Structural Engineers Association of California.
Bommer, J. and Martinez-Pereira, A. [2000] Strong-motion parameters: definition, usefulness and
predictability, Proc. of the 12th World Conference on Earthquake Engineering, Auckland. New
Zealand, Paper no. 206.
Bommer, J., Elnashai, A., and Weir, A. [2000] Compatible acceleration and displacement spectra for
seismic design codes, Proc. of the 12th World Conference on Earthquake Engineering, Auckland,
New Zealand, Paper no. 207.
Bommer, J. and Alarcn, J. [2006] The prediction and use of peak ground velocity, Soil Dynamics
and Earthquake Engineering 10(1), 131.
Booth, E. [2007] The estimation of peak ground-motion parameters from spectral ordinates,
Journal of Earthquake Engineering 11, 1332.
Calarasu, E. A. [2012] Microzonarea conditiilor locale de teren pentru municipiul Bucuresti si
zona sa metropolitana cu aplicatii in evaluarea hazardului seismic, Ph.D. thesis (in Romanian),
Technical University of Civil Engineering Bucharest, Romania.
Clough, R. and Penzien, J. [2003] Dynamics of Structures, 3rd ed., Computers & Structures Inc.,
Berkeley, California.
Elnashai, A. and Di Sarno, L. [2008] Fundamentals of Earthquake Engineering, John Wiley & Sons
Ltd., Chichester, West Sussex.
European Strong Motion Database. Retrieved from http://www.isesd.hi.is.
Fajfar, P., Vidic, T., and Fischinger, M. [1990] A measure of earthquake ground motion capacity to
damage medium-period structures, Soil Dynamics and Earthquake Engineering 9(5), 236242.
Itaca - Italiana Accelerometric Archive. Retrieved from http://itaca.mi.ingv.it/ItacaNet/.
Lungu, D., Coman, O., Cornea, T., Demetriu, S., and Muscalu, L. [1993] Structural response spectra to different frequency bandwidth earthquakes, Proc. of the 6th International Conference on
Structural Safety and Reliability, Vol. 2, Innsbruck, Austria, pp. 21632170.
Lungu, D. and Cornea, T. [1987] Power and Response Spectra in Bucharest for Vrancea Earthquakes,
Technical University of Civil Engineering Bucharest, Bucharest.
Lungu, D., Cornea, T., Aldea, A., and Zaicenco, A. [1997] Basic representation of seismic action,
in Design of Structures in Seismic Zones: Eurocode 8 Worked Examples, ed. D. Lungu, F.
Mazzolani, and S. Savidis (TEMPUS PHARE CM Project 01198: Implementiong of structural
Eurocodes in Romanian civil engineering standards, Bridgeman Ltd., Timisoara, Romania), pp.
160.
Lungu, D., Demetriu, S., and Cornea, T. [1992] Frequency bandwidth of Vrancea earthquakes and
the 1991 edition of seismic code in Romania, Proc. of the 10th World Conference on Earthquake
Engineering, Vol. 10, Madrid, Spain, pp. 56335638.
Newmark, N. and Hall, W. J. [1969] Seismic design criteria for nuclear reactor facilities, Proc.
of the 4th World Conference on Earthquake Engineering, vol. 2, Santiago de Chile, Chile, pp.
B5.112.
Newmark, N. and Hall, W. J. [1982] Earthquake Spectra and Design, Earthquake Engineering
Research Institute, Berkeley, California.
PEER Strong Motion Database. [2006] Code for seismic design Part I Design prescriptions
for buildings, Ministry of Transports, Constructions and Tourism. Retrieved from http://peer.
berkeley.edu/peer_ground_motion_database.P100-1/2006
Rathje, E., Abrahamson, N., and Bray, J. [1998] Simplified frequency content estimates of earthquake ground motions, Journal of Geotechnical and Geoenvironmental Engineering 124(2),
150159.
Strong ground motion database of Turkiye. http://kyhdata.deprem.gov.tr/2K/kyhdata_v4.php.
Wald, D., Quitoriano, V., Heaton, T., and Kanamori, H. [1999] Relations between peak ground
acceleration, peak ground velocity, and modified Mercalli intensity in California, Earthquake
Spectra 15(3), 557564.
559
Yakut, A. and Yilmaz, H. [2007] Evaluation of spectral ground motion intensity parameters,
Proc. of the 8th Pacific Conference on Earthquake Engineering, Singapore, Singapore, Paper
no. 036.
Zhu, T. J., Tso, W., and Heidebrecht, A. [1988] Effect of peak ground A/V ratio on structural
damage, Journal of Structural Engineering 114, 10191037.