EARTHQUAKE WAVE PROPAGATION COMPARATION ANALYSIS IN

SHAKE2000AND PLAXIS V8 BASED ON SNI 03-1726-2012 FOR NORTH, CENTER, AND

SOUTH JAKARTAREGION

Eng Phin Binus University, Jl. KH. Syahdan No. 9 Kemanggisan Jakarta Barat, (+62)87881612401,

engphin@yahoo.com Eng Phin, Gouw Tjie Liong

ABSTRACT

Response Spectra Analysis using SHAKE2000 and Plaxis v8 was meant to understand the difference in wave propagation induced by difference in models and required parameters. The analysis was conducted by modelling the soil at five locations in Jakarta, and by using synthetic ground motion for 2500 year period. Analysis result shows that in general, site specific spectral values obtained are larger than that from SNI 03-1726-2012. Besides that, the result from Linear Elastic Model (Plaxis) gives relatively very high value compared to that of the other models, and is not suitable for seismic analysis. Mohr Coulomb model gives relatively high result, and less suitable for seismic analysis. Hardening Soil with Small Strain Model is more suitable for seismic analysis. Linear Equivalent model (SHAKE2000) gives closest result to SNI compared to that of other models. But, to achieve even better results, more detailed study regarding seismic parameter determination for soils in Jakarta along with soil investigation procedure for seismic analysis would be necessary. Keywords: Response Spectra, SHAKE2000, Plaxis, SNI 03-1726-2012, Jakarta

INTRODUCTION

In understanding earthquake phenomenon, one big problem that often occurs is the difficulty in analyzing by using conventional ways, especially when the soil at the particular site varies greatly. If the soil types varies greatly and have different thickness, then conventional calculation to acquire the response of soil surface because of earthquake would become so ineffective and inefficient. Rapid development in recent days, especially in Information and Communication Technology has become a stepping stone for researchers to conduct study regarding earthquake. The presence of applications that are capable of conducting dynamic analysis makes the calculation becomes far more effective and efficient. In this research, Plaxis v8 and SHAKE2000 will be used to analyze the response of the surface of soil caused by earthquake. Earlier study that relates to Dynamic Analysis using Finite Element Method shows that the choice of soil model type (Mohr Coulomb, Hardening Soil Small Strain, etc) will affect the result obtained. Herold et al (2009) has conducted the research regarding deformation in soil caused by dynamic loading using Mohr Coulomb and Hardening Soil Small Strain. The result from the research shows that modelling using HS Small gives a result that matches closer to that from field observation, compared to the result obtained from Mohr Coulomb Model. And also, a research done by Rafal (2010) also shows that HS Small gives a result that closely matches that from field observation. In this research, modelling scope will be widened, by comparing the result obtained from Finite Element Method modelling (Mohr Coulomb, HS Small, and Linear Elastic) to the result from the most popular wave propagation analysis method, which is the Linear Equivalent, by using SHAKE2000, and also by comparing to the result from experts. To simplify the studied problem, then it is necessary to limit the research area in this research, which are: • The location of the soil is from North Jakarta, Central Jakarta, and South Jakarta. • The analysis is conducted by using Plaxis v8.2 and SHAKE2000. • The results that will be compared are the Acceleration Design Spectra.

• As a comparation, SNI 03-1726-2012 will be used. The objective of this research is to compare the results from earthquake wave propagation analysis generated by Plaxis v8 and SHAKE2000, by forming Response Spectra from acceleration output at soil surface, and analyzing what makes the differences in result. The benefit of the research is to understand how far the difference in the output generated by those two softwares, and understand the causes, therefore those could become some considerations when using those softwares.

METHODOLOGY

Methodology and process in this research is as stated below: a. Literature study and gathering secondary data, which are soil investigation reports for North

Jakarta, Central Jakarta, and South Jakarta from the officials or institutes that owned geotechnical data for Jakarta Region.

b. Carry out data selection, to determine which geotechnical datas that are valid to be used for wave propagation analysis from the base rock to soil surface. Some aspects to be considered are: • Information about soil testing location (North, Central, or South Jakarta) • Soil layer data, in the form of soil type description, and the thickness of every soil layers. • N-SPT value and Shear Wave Velocity information for every soil layer. • Laboratory test results in form of soil unit weight, Plasticity Index.

c. Determine the depth of base rock. Base rock depth can be determined by using earlier studies done by Asrurifak, M. et al (2013) in a journal which title is “Pengembangan Peta Klasifikasi Tanah dan Kedalaman Batuan Dasar untuk Menunjang Pembuatan Peta Mikrozonasi Jakarta Dengan Menggunakan Mikrotremor Array”. Base rock depth is determined by using a contour map for base rock depth, and by plotting project location to a map application (Google Map, etc), the base rock depth could be determined.

d. Carry out linear interpolation for the depth intervals which are not tested. The soil is assumed to be Clay, which at base rock border, the Shear Wave Velocity is assumed 750 m/s. Then, for the unit weight of soil, start from the end of borehole test until base rock is 19 kN/m3.

e. Create the soil model. Modelling is conducted for four types of soil model, which are Mohr Coulomb, HS Small, Linear Elastic, and Linear Equivalent. For the first three models (Mohr Coulomb, HS Small, Linear Elastic) modelling will be done by using Plaxis, and for Linear Equivalent Model, SHAKE2000 will be used.

f. Inserting the time history data in the form of base rock acceleration caused by earthquake, which is obtained from seismic hazard analysis result for Jakarta region, and with 2% probability of exceedance for 50 years return period ((≈ 2500 years).

g. Carry out analysis using Plaxis and SHAKE2000. After that, forming Response Spectra based on the analysis result from Plaxis and SHAKE2000. The Response Spectra then further processed to become Recommended Spectra, which finally will be compared to Design Spectra SNI 03-1726-2012, so the cause of the differences can be seen, and important aspects to be considered can be determined.

HASIL DAN BAHASAN

The result of the research obtained is in the form of Response Spectra graph, which is the graph depicting structure response toward earthquake vibration at the surface of soil. Below presented are the graphs for five studied location, by using five types of earthquake.

Cempaka Putih Pantai Mutiara s

Figure 1 Cempaka Putih Analysis Result Figure 2 Pantai MutiaraAnalysis Result

Cilandak Ancol

Figure 3 Cilandak Analysis Result Figure 4 Ancol Analysis Result

Kuningan

Figure 5 Kuningan Analysis Result

Based on obtained analysis result, it can be seen that in general way, spectral value from Site Specific analysis gives greater value than SNI, including spectral value obtained from the study carried out by expertise. This shows that using the values obtained from site specific analysis for designing will give a more conservative design result. But on the other side, site specifc analysis basically includes some assumptions in it. Significant difference in spectra result from SNI and Site Specific can be caused by insufficient study regarding deeper soil layer (>100 m), while usually soil testing with borehole only reaches 60-90 m. For deeper soil modelling until it reaches base rock, typically linear interpolation will be used. For example, shear wave velocity, Vs, which testing only reaches 30 m depth, while the modelling could reach 500-700 m depth. Base rock depth also comes with its own problem. Up until now, the research regarding base rock depth is still insufficient. The deeper the base rock, typically the earthquake that reaches surface will be smaller. And if there is an error in determining the base rock depth, then the result of the analysis would also be affected. Besides that, small strain parameters (G/Gmax and Damping Ratio) for soils in Jakarta remains unknown. Parameter determination is conducted by matching soil classification and Plasticity Index obtained with curves that have been published. And there are some general parameters, such as elasticity modulus, E, cohesion, c, and internal angle friction, φ,which are usually not tested when carrying out Seimic Tests. This is because mostly used soil model, which is Linear Equivalent, do not take into account for large strain and failure condition, which is different from Finite Element Method (2 Dimension). Therefore, when modelling using Finite Element models, some correlations would be necessary. Correlation chosen will affect the result obtained. Generally speaking, spectra value generated by Linear Equivalent modelling in SHAKE2000 is lower than that of any other models, but higher than that of SNI. It could be concluded that SHAKE2000 still gives an acceptable result, and the result also matches quite well with the result obtained from the expert. The spectral value obtained from analysis using Linear Elastic modelling gives a relatively very high value, far above if compared to the results obtained from another modelling. This very high spectral value is caused by some factors, as stated below: • When failure occurs in the soil, earthquake wave energy will be absorbed into deformation. But,

in Linear Elastic modelling, soil cannot experience any failure, so earthquake energy that propagates through soil body until the surface will not decrease.

• The model only uses single value of soil stiffness, that is Eref, which is taken from the initial stiffness value, Emax. Therefore, soil behavior in small strain cannot be depicted in the model.

It can be said, that Linear Elastic model is not suitable for seismic analysis. For Mohr Coulomb model, it can be seen that spectral value generated is quite high, but lower than that of Linear Elastic. High spectral value from Mohr Coulomb model comes from soil behavior in the model, where the soil acts linearly as long as it has not reach failure state. This is different from soil behavior under small strain condition, where even in small strain, plastic deformation still occurred. And when the soil reaches failure, plastic deformation takes place, causing the earthquake energy reduces. Therefore, it can be concluded that Mohr Coulomb is less suitable for seismic analysis. If it is needed, then the result obtained would become overestimated. Modelling using Hardening Soil Small Strain gives quite good result. Generally, the result obtained from this type of model gives lower value than that of Mohr and Linear Elastic modelling. Besides that, from the parameters used, it can also be seen that this model is capable of taking into account the soil behavior under small strain. But, to ensure a better result, then larger strain parameters determination should also be considered. This condition is one difficulty when adopting the model for analysis. Finally, based on the result of the research, it can be said that for seismic analysis conducted for perfectly horizontal soil layers, SHAKE2000 with its Linear Equivalent model gives a result that matches better with SNI. One of some aspects that caused it is because analysis carried out in Indonesia until now still use Linear Equivalent analysis. But, this conclusion is still temporary, and further and deeper studies would be necessary.

CONCLUSIONS AND SUGGESTIONS

Based on analysis result that has been done, some conclusions are obtained, as stated below: • The difference in Response Spectra analysis result compared to SNI is caused by some points.

First, the study regarding deeper soil layer parameters (>100 m) up until now can be said still insufficient. And also, base rock depth still cannot easily determined. Besides that, small strain parameters for soils in Jakarta still remains unknown.

• Based on analysis result, Linear Elastic soil model gives a much higher value compared to the result obtained from another models. It can be concluded that Linear Elastic soil model is not suitable for seismic analysis.

• Analysis result shows that spectral value obtained from Mohr Coulomb model is quite high compared to that from Linear Equivalent model (SHAKE2000). Mohr Coulomb soil model is less suitable for seismic analysis. Can be used, however it gives overestimated result.

• Hardening Soil Small Strain soil model is quite good for seismic analysis. This can be seen from the spectral value that matches well with the result from SHAKE2000.

• SHAKE2000 (Linear Equivalent) soil model gives closer result to SNI compared to Hardening Soil Small Strain, because basically, analysis performed in Indonesia to generate response spectra for SNI still uses Linear Equivalent soil model like the one in SHAKE2000.

There are also some suggestions that can be considered for further studies, as below: • Further studies regarding the compatibility of the result from SHAKE2000 for Jakarta seismic

conditions is needed. • Regarding the depth of testing for seismic analysis, such as Borehole, SPT, and Seismic

Downhole, an increase in the depth would be better. • It is also adviseable to do further studies that includes Rayleigh Damping in Plaxis.

REFERENSI

Agus. (2013). Penggunaan RSNI 03-1726-201X dalam Perancangan Struktur Gedung Tahan Gempa di Kota Padang dan Perbandingannya dengan SNI 03-1726-2002. Jurnal Momentum. 14 (1): 11-17.

Asrurifak, M., et al. (2013). Pengembangan Peta Klasifikasi Tanah dan Kedalaman Batuan Dasar untuk Menunjang Pembuatan Peta Mikrozonasi Jakarta dengan Menggunakan Mikrotremor Array. Pertemuan Ilmiah Tahunan – XVII HATTI. 17(7).

Baker, J. W. (2008). An Introduction toProbabilisticSeismic Hazard Analysis. Brinkgreve, R. B. J. (1999). Beyond 2000 in computational geotechnics: Ten years of PLAXIS

international 1999. Rotterdam: Balkema. Chopra, Anil K. 2007. Dynamics of structures: theory and applications to earthquake engineering.

Upper Saddle River, N.J.: Pearson/Prentice Hall. Elnashai, Amr S., and Luigi Di Sarno.(2008). Fundamentals of earthquake engineering . Chichester,

U.K.: Wiley. Hutapea, B. M., Imanuel, M. (2009). Analisis Hazard Gempa dan Usulan Ground Motion pada

Batuan Dasar untuk Kota Jakarta. Jurnal Teknik Sipil:Jurnal Teoretis dan Terapan Bidang Rekayasa Sipil. 16 (3): 121-132.

Joner, W.B. and Boore, D.M. (1992). Predictions of earthquake response spectra. Proc. 51st Annual Convention of Struct. Eng. Assoc. of Calif.; also USGS Open File Report 82-977.

Kramer, Steven Lawrence. 1996. Geotechnical Earthquake Engineering. Upper Saddle River, N.J.: Prentice Hall.

Merati, W., Masyhur I. (2000). Development of Synthetic Ground Motions – Bedrock of Jakarta. Twelfth World Conference on Earthquake Engineering, New Zealand.

Nigam, N. C., Jennings, P. C., & California Institute of Technology. (1968). Digital Calculation Of Response Spectra from Strong-Motion Earthquake Records. Pasadena, California: California Institute of Technology, Earthquake Engineering Research Laboratory.

SNI 03-1726-2012, Tata Cara Perencanaan Ketahanan Gempa untuk Struktur Bangunan Gedung dan Non Gedung.

World Conference on Earthquake Engineering, & New Zealand National Society for Earthquake Engineering. (2000). 12 Wcee 2000: 12th World Conference on Earthquake Engineering. New Zealand: New Zealand National Society for Earthquake Engineering.

Herold, Andreas. (2009). The Use of Hardening Soil Model with Small-Strain Stiffness for Serviceability Limit State Analysis of GRE Structures.

RIWAYAT HIDUP Eng Phin lahir di kota Jakarta pada 06 Agustus 1992. Penulis menamatkan pendidikan S1/S2/S3 di Universitas Bina Nusantara dalam bidang Teknik Sipil pada tahun 2014.