Identification of liquefaction potential using empirical and numerical approach on Maranatha Area, Sigi Regency

. A Liquefaction that hit Palu City and Sigi Regency, Central Sulawesi on September 28th, 2018 triggered by a 7.5 magnitude earthquake has been stated as National Disaster by BNPB. This paper aims to identify the liquefaction potential considering the soil structure, N-SPT data, seismic history, water level conditions, and relevant geotechnical data at Maranatha Area. This Village located very close to the liquefaction occurance mentioned above is chosen as the main focus location of this research. The research focused on Maranatha section of Gumbasa Irrigation Rehabilitation Project-3 rd Package. Eight selected locations along the project are analyzed using the Simplified Method to identify the liquefaction potential. Based on geotechnical investigation found that the soil conditions of some areas are cohesionless. Soil exploration shows that most soil deposit consists of sandy and clay layers from the ground surface up to a depth of 20,00 meters, and a high elevation of groundwater level ( -2.10 to -9.00). The potential analysis of liquefaction based on numerical approach empowers the empirical approach which indicates that among 8 boreholes, 3 of them were quite potential for liquefaction to occur and Borehole-37 was to be on concern due to its potential for flow liquefaction. It is found that certain depths of soil layers of the areas are prone to liquefaction with Potential Index of 1.70 to 10.55 (Low to High Potential).


Introduction
The liquefaction Phenomenon in Palu City and its surroundings that occurred on September 28th, 2018 had been stated as National Natural Disaster by Indonesian National Disaster Relief Agency (BNPB) and spotlighted in both National and International Media.A quite significant earthquake of 7,5 Moment Magnitude which is identified as a shallow earthquake classification (10 km below the surface) has been causing liquefaction in several areas near Palu City [1][2].
The quake devastated the entire city and derived other hazards such as tsunamis and liquefaction.Most of the area in Palu City is classified as High Liquefaction Vulnerability Zone.Proven by real cases and also strengthened by the Indonesian Earthquake Zoning Based on Indonesian Liquefaction Vulnerability Zone published by Geology Agency, Ministry of Energy and Mineral Resources [3] Earthquake Map by National Earthquake Study Center Team [4] which also classifies Palu City and its surroundings as an earthquake-prone area, supports the hypothesis that there may be a repetition of liquefaction in Palu City and its surroundings.
On this basis, this paper derived a case study of the Gumbasa Irrigation Channel Reconstruction and Rehabilitation Project at Marantha Area (Fig. 1).This case is chosen due to absence of liquefaction potential analysis on this area.The area of this project is located 15 km from Palu City to the Southeast.Due to its close distance from previous liquefied sites, a study needs to be done to ensure that the project is safe from Liquefaction.A recent hypothesis is constructed that the location has a high chance of liquefaction, as the distance is near and the visual layer of soils is similar to the liquefied sites.Although later then found the area is prone to liquefaction, a mitigation plan could be prepared at an early stage to prevent the damages that may happen.Eight Boreholes (BH 30 to BH 38) were carried out which then be analyzed in determining the liquefaction potential.The distance between each borehole is about 300 meters away from each other.The total distance of this focus is 2.4 kilometers starting from BH-30 to BH-37.A numerical approach will also be done to know the mechanism of liquefaction which may be occurred near the sites.

Sets of data
The data which will be used in this study would be primary and secondary.The list of used data are including geotechnical data such as N-SPT, Seismic data, Liquefaction and Earthquake mapping zone, Geophysical Investigation data, and Groundwater level.The SPT data is the most crucial since it was the main parameter to determine liquefaction potential analysis using the Simplified Procedure [6].
The SPT data will be supplied by BWS Sulawesi III.The Deterministic approach to pointing the PGA will be calculated using the chosen attenuation formula [7].Some groundwater levels will be obtained in one package with the Borelog data dan some of the data is primarily taken.A geological map presents a variety of information on rock types and distribution, structure, morphology and slope, soil susceptibility, and sequences of rock variations.This kind of map is indispensable for making regional development decisions, especially for strategic projects.The current project is located in Sigi Regency, Central Sulawesi (Fig. 1).Fig. 1.Study area (Maranatha Section) [5].
Meanwhile, The regional geological map is obtained by accessing Sigi Official Government's Website at bpbd.sigikab.go.id.The MASW data are used to determine the site classification and empower the interpretation of SPT data.The MASW data are supplied by Balai Wilayah Sungai Sulawesi III, Directorate General of Water Resources.
A bore log is a record that describes the in-situ circumstances and elements of geotechnical exploration activities, such as mining and drilling boreholes.Visual inspection of samples acquired from the geotechnical process yields the log data required to generate a borehole log.Relevant data from trusted sources.In this research, the bore log data is secondary data received by BWS Sulawesi III on Gumbasa Rehabilitation and Reconstruction Project.

Liquefaction potential identification method
In this research, a simplified method [6] will be used to determine the potential of the liquefaction phenomenon.The data needed on this formula is the PGA data (can be determined by Probabilistic or Deterministic Approach), N-SPT Value data, Bore-logs, and the soil properties data.
A research was conducted related to a comparative study of several liquefaction potential analysis methods by testing them on liquefaction cases on the coast of Bengkulu City in 2007 [8].The results of this study indicate that the Idriss and Boulanger method provides a result that appeared to be the closest approachment to the events at the research locus with the lowest level of error.
Susceptibility factor of liquefaction had been formulated [6] as a function of the ratio between Soil Resistance due to cyclic loading and the stress received by the soil due to cyclic loading.The value of resistance to cyclic loads is also known as Cyclic Resistance Ratio (CRR) which is calculated based on the N-SPT or Cone Penetration Test values.Liquefaction that occurs in saturated sandy soils can be caused by the seismic stress ratio (SSR) or commonly called the Cyclic Stress Ratio (CSR).
Cyclic Stress Ratio is dominantly caused by earthquakes.The value of CSR is affected by maximum earthquake acceleration, earth acceleration, total soil stress, effective soil stress, and depth factor reduction [6].A comparison of CRR and CSR values is defined as a safety factor value that implies soil susceptibility to liquefaction hazards (Equation 1).If the value of the safety factor (FS) is less than 1 then the soil layer is considered vulnerable to liquefaction potential.

Cyclic Resistance Ratio
On the basis of data from field soil strength tests, including N-SPT and CPT measurements, the cyclic resistance ratio (CRR) value was determined.Based on field-collected N-SPT data, an equation was developed to determine the CRR [9].The SPT value used is the value that has been corrected for the parameters related to the sampling technique, the tools used (Standard penetration test) on soils, and the content of fine grains.Formulated CRR for a specific earthquake of 7.5 MW as Equation 2.
( 1 ) 60 = ( 1 ) 60 + ∆( 1 ) 60 (3) ( 1 ) 60 =   .  .  .  .  .  (5) where The SPT value after correction is stated as (N1)60cs.The type of tool used during the NSPT penetration test affects the value of each corrective factor, where CN is the overburden correction factor, CE is the energy ratio correction factor, CB is the borehole diameter correction factor, CR is the rod length correction factor, and CS is the sampling correction factor.On the field equipment list, calculations for CB, CE, CN, CS, and CR are needed.The equipment adjustment in this instance is worth 1.216.

Cyclic stress ratio
Cyclic Stress Ratio represents the stress ratio caused by cyclic loads.CSR was first introduced in 1971 [9] and then developed in 2014 [6] in the form of Equation 6.
where  is the total stress on the soil,  is the shear stress reduction factor, max is the greatest horizontal acceleration at the ground surface caused by the earthquake, and ' is the effective stress at a depth of z meters.Only depths under 34 meters are allowed to utilize this equation.At the location of the selected boreholes, the computations were repeated for each stratum.

Attenuation
The author uses the attenuation equation developed in 2006 [7].The method chosen were demanding the criteria such as the depth of the earthquake, and the magnitude of the earthquake and also because this equation was found in Japanese studies which have similar cases of seismicity to Indonesia.The equation is divided into 2 formulas, which can be applied for shallow earthquake data with a depth of < 30 km and deep earthquake data with a depth of > 30 km.In this case, the formula used is as shown in Equation 7.
o  = 0.56w -0.0031X − log(X + 0.0055 × 100.5 × w) + 0.26 + 0.37 (7) where Mw is the earthquake's magnitude, X is its hypocenter's from the point of view of the earthquake in kilometers, and D is its depth (D 30 km).

Numerical model concept
Numerical modeling was carried out to crosscheck and see the behavior of the ground against the simulated earthquake.Soil parameters are inputted into the numerical software as a representation of the field conditions.In general, models are designed to answer specific questions.Several simplifications were made in this model considering soil parameters with a lot of uncertainty in it.In this study, Geostudio Quake/W finite element numerical model was chosen to simulate liquefaction events that might occur in the test location area.A geotechnical finite element software program called QUAKE/W is used for the dynamic study of earth structures that are shaken by earthquakes, the liquefaction process, and other abrupt impact loading such, for instance, dynamiting or pile driving.Due of its capability to depict the liquefaction phenomena, this program was selected to execute the simulation.The layer model and the parameters utilized were both simplified in the model.The simplification concept is shown in Fig. 2. The corrected value for Shear Stress Ratio is designated as Ka, which stands for alpha.The dynamic stress calculated by QUAKE/W can be considered as the corrected field stress (corrected field CSR).The userdefined Cyclic Number Function generally represents the ratio of field cyclic stress (CSR field) before correction.Therefore in QUAKE/W, the calculated CSR value with finite elements is divided by Ka to get the value that corresponds to the value specified in the Cyclic Number function in Equation 8.
The definition of ground surface collapse (Fig. 3) with static conditions optionally specified can be used QUAKE/W to mark an element as 'Liquefaction' [10].In simulating the liquefaction behaviour, some parameters were used in this model including b, Sat, Soil Type, Poisson's Ratio , Permeability and Modulus Elasticity.Boundary condition is needed to determined the model used in this case, based on numerical numeric.
Most frequently, nodal displacements are stated to provide a frame of reference for the analysis; typically, the displacement is zero.Consider the circumstance shown in Fig. 4 as an example.The displacement is indicated as zero at the issue's base.This indicates that the computed motion will be in relation to a fixed basis in both the x and y directions.
Although it is not exactly true to specify a vertical displacement of zero at the ends, the boundary is sufficiently removed from the slope for a zerodisplacement boundary to have little impact on the dynamic shear stresses in the slope, which is what this study is primarily trying to determine.

Liquefaction potential index
The developed research on liquefaction potential level is also known as liquefaction potential index (LPI) [11] to predict the level of liquefaction potential based on the thickness of the liquefaction potential soil layer and the distance of the liquefaction potential layer to the surface.The LPI value can only predict the potential for liquefaction to a depth of 20 m because the liquefaction effect at a depth of more than 20 m on the soil surface rarely occurs The LPI value is calculated by Equation 9.  = ∫ .()  20 0 (9) where F = 1-Fsliq for F < 1.0 and F=0 for FS liq > 1.0 and W(z) = 10 -0.5z; z is in meters.The LPI value the level of liquefaction potential in the area studied with a very low to very high potential level with the provisions of the liquefaction level shown in Table 1.
Table 1.Classification of liquefaction potential [11].Analysis of liquefaction potential can be strengthened by analyzing the possibility of liquefaction in the soil.The probability of liquefaction will show how likely it is potential of liquefaction will occur in a particular area.This probability analysis is intended only as an initial estimate for the possibility of liquefaction and helps add information on critical conditions Fsliq = 1.

Stratigraphy
Based on some N-SPT and bore log data received from BWS Sulawesi III, the interpreted stratigraphy (Fig. 5) shows that most of the layers around the Gumbasa Irrigation Area, especially the Maranatha Section, are sandy soil types.This reinforces the statement which mentions sandstone as rock formations around Palu City and Sigi Regency [12].Considering that Maranatha Village, which is located in Sigi Regency, is an area that is very prone to earthquakes [4], can be set an initial hypothesis that the location has a tendency for liquefaction to occur.It was indicate as previously stated due to its dominant layers of sand.The information on groundwater table (GWT) data as well as the density level of the soil needs to be investigated further to a certain level of potential liquefaction that may be occurred.

Liquefaction potential analysis based on empirical approach
The empirical approach in determining the liquefaction is considering the Fine Contents, PGA, and Shear Stress.The calculations were then recapped at the critical depth of each Borehole (Table 2).The critical depth was referred to the soil Layer which was found in several criteria; Sandy, Saturated, and at loose condition (Lower N-SPT).The layers that appeared to be non-Sandy layers (Clays, Gravels, etc) will not be calculated as the non-sandy layer has the least probability of liquefaction occurrence.Some of the groundwater levels at several points were being concerned due to the peak level of groundwater which then had a high chance of triggering liquefaction.The highest groundwater level is -2.1 (below the ground) at BH-37 and the lowest simulation of the groundwater level is -7.5 m below the ground surface at BH-31 (Table 2).By calculating using the condition of a 7.5 Mw earthquake, the results (Table 3) show 3 of 8 boreholes were found that certain layers of soil might occur liquefaction.
Table 3 is showing the value of the Safety Factor of each layer in every borehole.The result can be interpreted that 3/8 boreholes are vulnerable to liquefaction.Some of the boreholes consist of dominant layers of clay which are then classified as the nonliquefiable layer.As mentioned, liquefaction occurred if the value of FoS is below 1, in that term, 1.5 m layer of soil at BH-33, 4.5 m layer of soil at BH-35, and 10.5 m layer of soil tend to liquefy.Detail of calculations in defining liquefaction by comparing CSR and CRR, especially at its critical layer status are shown on Table 2 and Fig. 6.Fig. 6.CSR vs CRR at each borehole.

Liquefaction potential analysis based on numerical approach
As previously stated, a numerical approach was done to find the information of the liquefaction mechanism.The model was run in a dynamic module by using a 7.3 Mw Chi-chi Taiwan earthquake with a duration of 40 seconds due to the restriction of 7.5 Mw Palu earthquake data.The recent earthquake data was chosen due to its similarity with the previous devastating earthquake that occurred in 2018; both are strike dip type earthquakes, shallow crustal, and 40 seconds in duration.
Based on the Empirical result, it is known that Borehole-37 is the most prone point to liquefied, so in this case, BH-37 will be acted as a sample in numerical modeling.The model has been simulated, and it is found that the layer begins to liquefy at 4.37 sec after the earthquake occurred.The simulated earthquake duration is about 40 seconds.The simulation result (Fig. 7) shows that the middle point of the soil layers tends to be liquefied first (lower value of CSR) or to be precise, the liquefaction begins from the middle with a mechanism of spreading type to both the right and left side.With this kind of pattern, there is a possibility that flowliquefaction may also be occurred at Maranatha Section if the 7.3 Mw Earthquake hit the area.Time comparison between the liquefy time and the total earthquake duration is highly contrasted.By the predicted mechanism of flow liquefaction, it may lead to a devastating impact on the structure above the liquefied layers, which in this case is Gumbasa Irrigation Canal Rehabilitation Project and Existing structure of Pasigala Raw Water Transmission.

Mapping liquefaction potential index
Using the method of Liquefaction leveling [11], the liquefaction potential index at each borehole was determined.The result shows that among 8 Boreholes, 3 of it were showing the indication of the probability of Liquefaction occurrence.The recapitulation of the LPI value is shown in Table 2.
The risk at the borehole points observed is classified as Non-Liquefied, Low to High in Liquefaction.The potential risk is shown in Fig. 6.

Conclusion
Referring to the result and discussion section, it can be stated that the geological condition along the site consists of sandy and clay layers.The groundwater level from the Borelog data and primary observation is also found the GWL is under 10 meters which means the chance of liquefaction to occur should be aware.Among the 8 boreholes studied, 3 of them are prone to liquefaction at a certain depth and layer thickness by simulation using 7.5 Mw and 7.3 Mw strike slip earthquakes.Borehole-37 was found to be the most prone to liquefaction according to the Empirical and Numerical Approach.Numerical simulation, after the empirical approach was done, shows that flow liquefaction occurrence is probable at BH-37 (the concerned borehole due to high potential in liquefying) with 10.5 m liquefy soil thickness.It is found that certain depths of soil layers of the areas are prone to liquefaction with Potential Index of 1.70 to 10.55 (Low to High).The potential index indicates that some boreholes need to be concerned due to its "High Classification"; knowing that the project is still ongoing and some irrigation structures are still on construction.Further mitigation plans are needed to anticipate and lower the risk of liquefaction.
The Directorate General of Water Resources, Ministry of Public Works and Housing Indonesia under the name of Balai Wilayah Sungai Sulawesi III are acknowledged for their on going assistance, particularly in the data collection process.

Table 2 .
Liquefaction based on critical layer and depth (empirical result).

Table 3 .
The Factor of safety at boreholes.
n/a = not calculated due to non-liquefiable layer.

Table 2 .
Liquefaction potential index status.