International Journal of Engineering Research & Technology (IJERT)

ISSN: 2278-0181
Vol. 6 Issue 5, May - 2017

Simulation of Artificially Time History Load and Its

Impact on the Response of 2D Frame in Various Site
Classifications
Sandy I. Yansiku
Nusa Nipa University of Maumere

Abstract Dynamic time history approach as the most reliable method to observe structural response due to
earthquake loadings tends to be conducted by direct adoption of El Centro and Kobe earthquake records owing
to the lack of local time history accelerations data.
This study aims to engage program based simulations of various earthquake records with respect to
the provision of Indonesian standard using Seismomatch and Etabs to provide artificial seismic accelerations
for each site classification. A two-dimensional frame is analysed through response spectral and time history
techniques and compared the behaviour of the frame in terms of maximum story displacement, drift, stiffness
and moment.
Matching acceleration to target response spectrum using Etabs yields better average spectral curves
than that using Seismomatch. This, however, relies upon the scaling method and number of iterations.
Resulting story displacement, drift, stiffness and pier moment shows closely similar characteristics due to
dynamic loadings of original response spectral and its artificial counterpart. The utilisation of unmatched El
Centro and Kobe earthquake records considerably differs the maximum story displacement. The linear time
history analysis using Loma Prieta seismic record has resulted in extreme displacement of the proposed
structure for site B, while for site C, D and E the extreme response are using seismic records of Lucerne-z,
Saint Monica2 and Kocaeli respectively. Adopting appropriate seismic accelerations with matching earthquake
characteristics may lead to better structural analysis output.

Keywords: spectrum, matching, time history, displacement, behaviour.

I. INTRODUCTION specific seismic aspects to perform time history

Structures with extreme characteristics such as
vertical and horizontal irregularity could be hardly
analysed using static equivalent or response
spectrum method since these approaches cannot
exactly integrally demonstrate non-linear behaviour
of real ground motions. Time history method tends
to be the most appropriate and accurate technique to
estimate structure response due to dynamically
linear and non-linear seismic loadings [1].
Time history analysis requires earthquake
acceleration records of proposed structure location.
Despite the advantages of using original seismic
records, structural designers often deal with the lack
of sufficient strong motion records to meet the
seismic provision of the code [2]. Indonesian
standard, SNI 1726:2012 requires the minimum of
five records of horizontal ground motions with
analysis [3]. On the other hands, local earthquake
records with such characteristics may not be
currently available and hence direct utilization of
earthquake records with similar seismic
characteristics such as El Centro and Kobe appears
to be the only option for time history analysis.
Limits of local seismic data in Maumere, East
Nusa Tenggara province struggling with 6.8-SR
earthquake event in 1992 has led to the use of
merely El Centro earthquake records in time history
analysis. This condition limits the
comprehensiveness of dynamic time history
analysis either linear or non-linear [1].
This study aims to perform time history
matching simulation to generate average response
spectral and artificial time history acceleration for
dynamically linier response spectrum and time
history analysis for each local site classifications

1322
IJERTV3IS051552
 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 6 Issue 5, May - 2017

according to Indonesian seismic code. Moreover, Table 1 Spectral data for each site class
this study observes and compares structural Site Classification
Symbol
Rock B Hard C Medium D Soft E
response of a 10-story 2D frame in terms of story PGA (g) 0.446
displacement, drift, stiffness and maximum SS (g) 0.993
moment due to the resulting seismic curves and S1 (g) 0.402
CRS 1.092
accelerations between original records and artificial FA 1.000 1.003 1.103 0.909
acceleration. FV 1.000 1.398 1.598 2.400
SMS (g) 0.993 0.996 1.095 0.903
SM1 (g) 0.402 0.562 0.642 0.965
SDS (g) 0.662 0.664 0.730 0.602
II. METHODOLOGY SD1 (g) 0.268 0.375 0.428 0.643
A. Spectral Response T0 (detik) 0.081 0.113 0.117 0.214
TS (detik) 0.405 0.564 0.587 1.069
The proposed structure of this study locates in 8.000
Maumere, East Nusa Tenggara, Indonesia (Lat:- II
Ie 1
8.620842999999999, Long: 122.20770300000004). Several 0 3
seismic parameters were obtained from [4] dan Cd 5.5
0.0466
adjusted based on standard [3] for four site 0.9
x
classifications B, C, D and E as shown in Table 1. hn (m) 16,5
These parameters yield spectral response curves for Cu 1.4
Damping ratio 5%
each site class as shown in Fig.1. Seismic design category D

B. Earthquake Time History Records In comparison to Seismomatch, this simulation

SNI 1726:2012 stipulates that selected time uses Etabs for spectral matching although this
history records which consistently controls groud aplication is merely available for single record at
motions should be scaled such that time history one matching process adopting frequensi domain
response is close to the designed structural method. The method modifies Fourier amplitudo of
spectrum response [3]. Seismic acceleration a record based on the ratio of original spectral and
records used in this simulation includes 28 strong simulated spectral with fixed phase [7].
earthquake motion records with corrected duration
interval for data normalizing in matching process Table 2. Original acceleration records
using Seismomatch 2016. This software No. Event Symbol Duration Interval #Output
programatically calculate the difference and 1 ChiChi CHI 52.78 0.01 5278
iterates each accelerogram to obtain best matching 2 Friuli FRI 36.32 0.01 3632
3 Hollister HOL 39.93 0.01 3993
spektral with respect to the target spectral [5] for 4 Imperial Valley IMV 39.48 0.01 3948
each site with maximum difference of 15% and 5 El CentroNS ENS 31.18 0.02 1559
average maximum difference of 5%. Since 6 El CentroEW EEW 31.08 0.01 3108
matching accelerograms requires certain scaling 7 El CentroUP EUP 53.78 0.02 2689
8 Kobe KOB 40.90 0.01 4090
method, this simulation adopts scaling technique 9 Kocaeli KOC 34.96 0.01 3496
integrating area under spectrum curve [6] such that 10 Lander LAN 48.09 0.01 4809
the resulting spectral curve is not less than the 11 Loma Prieta LOM 39.90 0.01 3990
12 Northridge NOR 39.88 0.01 3988
target spectrum within range from 0,2T to 1,5T. 13 Trinidad TRI 21.40 0.01 2140
14 Lucerne1 LU1 24.05 0.01 2405
15 LucerneZ LUZ 24.05 0.01 2405
16 Lacc North2 LA2 29.98 0.01 2998
17 New Hall2 NH2 29.98 0.01 2998
18 SMonica2 SM2 29.98 0.01 2998
19 SMonicaZ SM3 29.98 0.01 2998
20 Petrolia1 PE1 29.98 0.01 2998
21 Petrolia2 PE2 29.98 0.01 2998
22 PetroliaZ PEZ 29.98 0.01 2998
23 Sylmarff1 SY1 29.98 0.01 2998
24 Sylmarff2 SY2 29.98 0.01 2998
25 SylmarffZ SYZ 29.99 0.01 2999
26 Yermo1 YE2 39.98 0.01 3998
27 Yermo2 YE2 39.98 0.01 3998
28 YermoZ YEZ 39.98 0.01 3998
Fig.1. Target spectral for each site class
Spectral matching will yield 5 seismic
acceleration records with best matching
convergence for each site class. By obtaining these

1323
IJERTV3IS051552
 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 6 Issue 5, May - 2017

five records whose average spectral meet the
requirement, the matched acceleration can be used
to observe structural behaviour with linear time
history analysis.
C. Structural Configuration
Proposed structure in this study is a ten story 2D
frame as shown in Fig.2, story height of 3.5m with
span of 5m.
D. Material Property and Dimension
Material property and size of columns, beams
and shear walls are provided in Table 3.
E. Loadings
Applied loads includes self-weight (SW),
superimposed dead load (DL), live load (LL) as
shown in Table 4.
Story 1-9 2.5 kN/m2
Seismic loadings refers to Indonesian standard
SNI 1726:2012 [3]. The given notation for static
equivalent load is EQX, dynamic spectrum
response load is RESPX and dynamic time history
load is THX. Static equivalent analysis takes into
account building weight and loads in Table 4 and it
is assumed 30% live load applied to each story.
Load combination is shown in Table 5 for initial
analysis phase comprising EQX, RESPX and single
THX, whereas Table 6 shows load combinations for
the second phase involving only linear time history
analysis due to five best matching accelerations.
Several terms such as ms (matching Seismomatch),
me (matching Etabs) and e (Etabs) are generated for
ease nomenclature.

Table 5. Load combinations of 1st phase

Code 1st phase load combinations
041 1.3324DL + 1.3324SW + 1RESPXB + 1LL
042 1.3324DL + 1.3324SW + 1RESmsB + 1LL
043 1.3324DL + 1.3324SW + 1RESmeB + 1LL
044 1.3324DL + 1.3324SW + 1THXeKOBEB + 1LL
045 1.3324DL + 1.3324SW + 1THXeELCENTROB + 1LL
051 1.3324DL + 1.3324SW + 1RESPXC + 1LL
052 1.3324DL + 1.3324SW + 1RESmsC + 1LL
053 1.3324DL + 1.3324SW + 1RESmeC + 1LL
054 1.3324DL + 1.3324SW + 1THXeKOBEC + 1LL
055 1.3324DL + 1.3324SW + 1THXeELCENTROC + 1LL
061 1.3324DL + 1.3324SW + 1RESPXD + 1LL
062 1.3324DL + 1.3324SW + 1RESmsD + 1LL
063 1.3324DL + 1.3324SW + 1RESmeD + 1LL
064 1.3324DL + 1.3324SW + 1THXeKOBED + 1LL
065 1.3324DL + 1.3324SW + 1THXeELCENTROD + 1LL
071 1.3324DL + 1.3324SW + 1RESPXE + 1LL
Fig.2. Proposed 2D frame
072 1.3324DL + 1.3324SW + 1RESmsE + 1LL
Table 3. Material property 073 1.3324DL + 1.3324SW + 1RESmeE + 1LL
Parameter 074 1.3324DL + 1.3324SW + 1THXeKOBEE + 1LL
Concrete strength, fc 25 MPa 075 1.3324DL + 1.3324SW + 1THXeELCENTROE + 1LL
Concrete unit weight, c 24 kN/m3 081 1.3324DL + 1.3324SW + 1THXELCENTROB + 1LL
Elastic modulus of concrete, Ec 23500 MPa 082 1.3324DL + 1.3324SW + 1THXELCENTROC + 1LL
Poisson ratio 0,2 083 1.3324DL + 1.3324SW + 1THXELCENTROD + 1LL
Beam B1 and B2 400mm x 600mm 084 1.3324DL + 1.3324SW + 1THXELCENTROE + 1LL
Column K1 700mm x 700mm 031 1.3324DL + 1.3324SW + 1EQXB + 1LL
Slab thickness 120 mm
032 1.3324DL + 1.3324SW + 1EQXC + 1LL
Shear wall thickness 250 mm
033 1.3324DL + 1.3324SW + 1EQXD + 1LL
034 1.3324DL + 1.3324SW + 1EQXE + 1LL
Table 4. Dead load and live load
Load type Unit
Dead load: The observation of structural behaviour is
Beam self weight 24 kN/m3 divided into two parts as previous description for
Slab weight 24 kN/m3 Table 5 and Table 6. In addition to that, the response
Waterproofing 2 cm at roof 0.14 kN/m2
Plafond 0.18 kN/m2 of the 2D frame can be observe based on the highest
Mechanical Electrical inst. 0.25 kN/m2 or the extreme structural response since this study
Specie (2 cm) 0.21 kN/m2 uses less than 7 records for each site class [8].
Tile (1 cm) 0.24 kN/m2
Live load:
Roof 1 kN/m2
Table 6. Load combinations of 2nd phase
1324
IJERTV3IS051552
 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 6 Issue 5, May - 2017

Code 2nd phase load combinations

046 1.3324DL + 1.3324SW + 1mECHIB + 1LL Table 7. Convergence of matching by Seismomatch
047 1.3324DL + 1.3324SW + 1mEFRIB + 1LL Target
Mean Spectrum
048 1.3324DL + 1.3324SW + 1mELOMB + 1LL Convergence #Iterations Ave. M ax.
Spectrum
difference difference
049 1.3324DL + 1.3324SW + 1mELUCZB + 1LL
ChiChi 16
050 1.3324DL + 1.3324SW + 1mELAC2B + 1LL Friuli 12
056 1.3324DL + 1.3324SW + 1mEFRIC + 1LL Loma Prieta 7
Rock B 2.41% 12.73%
057 1.3324DL + 1.3324SW + 1mELOMC + 1LL Lucerne1 11
058 1.3324DL + 1.3324SW + 1mELAC2C + 1LL LucerneZ 6
Lacc North2 8
059 1.3324DL + 1.3324SW + 1mENEWH2C + 2LL
Friuli 26
060 1.3324DL + 1.3324SW + 1mELUCZC + 1LL Loma Prieta 13
066 1.3324DL + 1.3324SW + 1mECHID + 1LL Hard C Lacc North2 15 3.07% 11.35%
067 1.3324DL + 1.3324SW + 1mEMON2D + 1LL New Hall2 28
068 1.3324DL + 1.3324SW + 1mEMONZD + 1LL LucerneZ 10
ChiChi 8
069 1.3324DL + 1.3324SW + 1mEPET1D + 1LL
SMonica2 17
070 1.3324DL + 1.3324SW + 1mESYL1D + 1LL Medium D SMonica3 25 3.06% 8.08%
076 1.3324DL + 1.3324SW + 1mEFRIE + 1LL Petrolia1 22
077 1.3324DL + 1.3324SW + 1mEKOCE + 1LL Sylmarff1 16
078 1.3324DL + 1.3324SW + 1mETRINE + 1LL Friuli 17
Kocaeli 4
079 1.3324DL + 1.3324SW + 1mELEX1E + 1LL
Trinidad 23
080 1.3324DL + 1.3324SW + 1mEPET2E + 1LL Soft E 2.23% 11.48%
Petrolia2 26
081x 1.3324DL + 1.3324SW + 1mESYL1E + 1LL Sylmarff1 20
Yermo2 16

III. RESULTS AND DISCUSSION

A. Matching by Seismomatch
Fig.3 shows the simulation output of 5
acceleration records in the form of response
spectrum curve and the target response spectrum
for each site classification. Best matching
convergence is shown in Table 7 as results of
certain number of iterations adopting method
proposed by [6] until reaching the boundary
condition and lowest difference. The average of
these five spectrums is scaled by 1.15 obtained
from the ratio of area under spectrum curve
between the matched and target spectrum along
0.2T to 1.5T. As can be seen, the red-dashed
average spectrums are above the target spectrums
along the specified range. As the matching
spectrums have met the requirement, the adopted
acceleration records can be used in dynamic time
history analysis.

B. Matching by Etabs
Since Etabs only provide single matching
process without input of scale factor, the similar
scale factors are inputted and calculated in
spreadsheet. Fig.4 depicts the matching results of
this application and Table 8 shows the difference
between target spectrums and scaled spectrums. It
can be seen in Fig.5 that spectral matching using
Etabs results in better average spectrums and closer Fig.3. Matched spectrums by Seismomatch
to target spectrums than using Seismomatch
although similar scale factor are applied to both The results, however, rely very much on scaling
process and Seismomatch provides scaling input method. Thus, it might be necessary for
facility for user to obtain better convergence value.
1325
IJERTV3IS051552
 International Journal of Engineering Research & Technology (IJERT)
ISSN: 2278-0181
Vol. 6 Issue 5, May - 2017

Seismomatch user to engage more number of Fig. 5. Scaled spectrum using Seismomatch and Etabs
iterations and trial more scaling factors for better
results.

C. Structural behaviour due to dynamic load of

average response spectrum
The averaged response spectrums are used to
observe structural behaviour. In Fig.6, the
maximum story displacement due to averaged
spectrums differs insignificantly for site class B
(3%-7%), whereas for site class C, D and E the
maximum displacement appears to be identic by
either Seismomatch or Etabs.
Moment of pier1 in Fig.7 shows similar
behaviour to the displacement. The difference of
pier moment due to the averaged spectrum and
target spectrum occurs only at story 1 to 5 for site
class B and D about 1% to 3%, while at story 6 to
10 there is no substantial difference for all site
classifications.
Large fluctuation of structural acceleration of
site B in Fig.9 appears to be due to the linear
integration during scaling process causing more
widened curve compared to other site classes.
In general, it can be summed up that matching
process yields comparable frame responses to the
original spectrum. Although the frequency domain
method cannot describe time series character of
seismic event which possibly increase total energy
of ground motions [7], the method in Etabs
generates closer spectrum response than
Seismomatch, which creates wavelets termed as
time domain method.

Table 8. Matching convergence using Etabs

Mean Spectrum
Target
Konvergensi Iterasi Rerata M aks.
Spektrum
perbedaan perbedaan
ChiChi 1
Friuli 1
Batuan_B Loma Prieta 1 23.44% 72.01%
LucerneZ 1
Lacc North2 1
Friuli 1
Loma Prieta 1
Tanah
Lacc North2 1 16.31% 44.09%
Keras_C
New Hall2 1
LucerneZ 1
ChiChi 1
SMonica2 1
Tanah
SMonicaZ 1 10.51% 47.42%
Sedang_D
Petrolia1 1
Sylmarff1 1
Fig. 4. Matched spectrums by Etabs Friuli 1
Kocaeli 1
Tanah
The relationship between story stiffness and Lunak_E
Trinidad 1 11.95% 7.44%
Petrolia2 1
story drift in Fig.8 shows similar characteristics.
Sylmarff1 1
The slight deviation at the first five stories is likely
owing to the lessen accelerations at these stories
compared to upper stories.

1326
IJERTV3IS051552
 International Journal of Engineering Research & Technology (IJERT)
Fig. 6. Maximum story displacement
Fig. 7. Moment of Pier1
D. Structural behaviour due to time history load
from single matching proces of El Centro and
Kobe records
Structural responses such as maximum story
displacement and pier moment are also observed
due to matched and unmatched El Centro and Kobe
earthquake acceleration using Etabs for each site
class.
Fig. 8. Story Drift vs. Stiffness
Fig. 9. Story Acceleration
IJERTV3IS051552
ISSN: 2278-0181
Vol. 6 Issue 5, May - 2017
In Fig.10, seismic loadings due to matched
accelerograms of both events generates similar
displacement, although unmatched Kobe
acceleration induces 31% less displacement than
unmatched El Centro acceleration. The matched El
Centro record generates 26% lower displacement at
the top story for site B then unmatched record,
whereas for site C, D and E the maximum
displacements at top story are 10%, 8% and 23%
greater respectively than unmatched record. The
use of matched Kobe records differs roughly 10%,
39%, 35% and 45% greater for all sites in that
order.
Matched El Centro acceleration induces similar
moment of pier1 compared to its original
acceleration for site C and D. Pier moment reduces
about 27% for site B when matched time series is
used and increase 7% for site E. Meanwhile,
matched Kobe has resulted in the rise of pier
moment of 6% - 10% for all site classes.
E. Structural behaviour due to time history load
from 5 matched records each site
Since five records of each site have met SNI
provision, structural responses due to seismic
loadings of these records can be observe in the
second phase using linier time history analysis.
The maximum displacement in x direction of
top story is 8.303 mm for site B caused by Loma
Prieta acceleration. The top story of structure on
site C, D and E displace about 10.322mm, 16.276
mm and 12.409 mm caused by dynamic loadings
of Lucerne-z, Saint Monica2 and Kocaeli seismic
events respectively as depicted in Fig.12.
Similarly, the maximum moment at top location of
Pier1 for site B, C, D and E are caused by Loma
Prieta, Lucerne-z, Saint Monica2 and Kocaeli in
that order.
Fig. 10. Maximum story displacement
1327
 International Journal of Engineering Research & Technology (IJERT)
Fig. 11. Moment of Pier1
Fig. 12. Maximum displacement of 2nd phase
Fig. 13. Moment of Pier1 in second phase
When comparing the results of this phase to the
previous section of time history analysis, it is
graphically obvious that pier moment differs
significantly as shown in Fig.13. In structural
analysis taking into account time history records, it
is profound to select correct, appropriate and
sufficient time history records since the possible
difference may affect the final design of the 2D
frame.
IV. CONCLUSION
Spectral matching with frequency domain
method in Etabs generates better average spectrums
by similar scaling factor than wavelet technique in
Seismomatch for all site classifications. This,
however, relies upon scaling techniques and
number of trials.
IJERTV3IS051552
ISSN: 2278-0181
Vol. 6 Issue 5, May - 2017
Dynamically earthquake loadings due to
average spectrums using Seismomatch and Etabs
yields comparable maximum displacement, drift,
pier moment and stiffness characteristics of frame
to the original spectrums of all sites.
Direct using of El Centro and Kobe records in
time history analysis varies structural displacement
and pier moment in all sites. Matched El Centro
records increases pier moment of site E, remains
stabled of site C and D and significantly decreases
pier moment of site B. Matched Kobe records
increases pier moment of all sites.
Four earthquake records of Loma Prieta,
Lucerne-z, Saint Monica2 and Kocaeli generate
extreme frame response in terms of displacement
and pier moment for site class B, C, D and E
respectively. The selection of appropriate and
sufficient earthquake records may produce ideal
artificial earthquake accelerations for both linear
and non-linear time history analysis.
V. RECOMMENDATIONS
Future study may simulate spectral matching
using time domain method in Etabs, which could
generate more accurate artificial time history
acceleration for non-linear analysis.
It is necessary to perform time history matching
response spectral prior to the time history analysis
as the yielding response of matched and unmatched
accelerations differs significantly.
VI. REFERENCES
[1] Huang, K., Minimum Number of Accelerograms for
Time-History Analysis of Typical Highway Bridges, in
The Department of Building, Civil and Environmental
Engineering 2014, Concordia University Montreal:
Quebec, Canada. p. 150.
[2] Fahjan, Y. and Z. Ozdemir, Scaling of Earthquake
Accelerograms for Non-Linear Dynamic Analyses to
Match The Earthquake Design Spectra, in World
Conference on Earthquake Engineering 2008: Beijing,
China.
[3] BSN, Tata Cara Perencanaan Ketahanan Gempa untuk
Struktur Bangunan Gedung dan Non Gedung. 2012,
Badan Standardisasi Nasional: Jakarta, Indonesia.
[4] PuskimPU. Nilai Spektral Percepatan Di Permukaan
Dari Gempa Risk-Targeted Maximum Consider
Earthquake Dengan Probabilitas Keruntuhan Bangunan
1% dalam 50 Tahun. Desain Spektra Indonesia 2011 17
Oktober 2016]; Available from:
http://puskim.pu.go.id/Aplikasi/desain_spektra_indones
ia_2011/.
[5] Seismosoft, Seismomatch. 2016.
[6] Al-Atik, L. and N.A. Abrahamson, An improved method
for nonstationary spectral matching. Earthquake
Spectra, 2010. 26: p. 601-617.
[7] CSI, ETABS Extended Three Dimensional Analysis of
Building Systems (User's Manual). 2010, Computers and
Structures, Inc.: California, U.S.A.
[8] ASCE, Minimum Design Loads for Buildings. 2005:
Reston, VA.
1328