Analysis of historical cases of liquefaction (sand boils) case study: Kelapa Gading Residence, Sigi Regency, Central Sulawesi Province

The Palu earthquake on September 28th, 2018, caused liquefaction in several locations in Central Sulawesi. According to The National Disaster Management Authority of Indonesia data, five prominent areas with massive impacts experienced liquefaction in flow liquefaction: Balaroa, Petobo, Jono Oge, Lolu, and Sibalaya. Meanwhile, the locations that experienced liquefaction in the form of sand boils needed to be recorded optimally. This research aims to see the parameters of soil layer behavior in areas that experience liquefaction in the form of sand boils. The study was conducted in Kelapa Gading housing, Sigi Regency, Central Sulawesi. Based on the initial survey, this location experienced sand boils in several areas. The research used the CPT-based liquefaction triggering procedure method and soil behavior index method. The result showed that the liquefaction potential was found at a depth of 2.5-3.3 meters for PGA 0.34 and -2.5-(-3.5) for PGA 0.68. When receiving cyclic load, the soil layer with an Ic value > 2.60 with a thickness of 1.6 m at the upper layer can withstand liquefaction with a thickness of 0.8 meters in the form of quicksand on the surface and only causes sand boils. Sand boils would not happen in the Kelapa Gading area if the excess pore water were redirected to discharge wells.


Introduction
Sulawesi Island is part of the Indonesia Archipelago.This island is the interconnection between Indian-Australian, Eurasian, Pacific, and Philippine Sea plates.The plates' interconnection creates deformation characterized by the presence of subductions and major regional strike-slip faults [1,2], which cause tectonic earthquakes-prone zone.The last profound quake on 28 th September 2018 at Palu, Central Sulawesi, with Mw 7.5, was believed to be an impact of the Palu-Koro fault strike-slip movement.Figure 1 shows the primary location of the Mw 7.5 earthquake event with the overall shake map of all other quakes within the area.The cyclic loading from the earthquake caused liquefaction.It caused devastating damage in several Palu City dan Sigi Regency locations, including Balaroa, Petobo, Jono Oge, Lolu, and Sibalaya [3][4][5].It was reported that the liquefaction alone caused at least 3720 houses and their supporting lifeline to be permanently shattered [6].
Liquefaction occurs when the saturated loose sand is induced by cyclic loading produced during the earthquake.The loads force sand particles to move, create normal stress, and then transfer to pore water.This causes a reduction in confining stress within the soil and provokes soil deformation due to its loss 1314 (2024) 012042 IOP Publishing doi:10.1088/1755-1315/1314/1/012042 2 of strength and stiffness [7].Liquefaction can be divided into flow and cyclic mobility (lateral spreading and sand boils) [8].Most recorded data and research on liquefaction in Palu dan Sigi focused on flow liquefaction, while this paper will focus on cyclic mobility in the form of sand boils.Sand boils occur vertically when excess pore water pressure from the saturated soil layer is released [8].One of the sand boils-affected sites was Kelapa Gading Residence.Kelapa Gading Residence is approximately 1km from a liquefaction-affected area in Petobo and is located near the Palu River (Figure 2).The area's ground level is also lower than Petobo's liquefied site(Figure 3).Due to its distance and strategic location, the liquefaction survivors selected this residence as a relocation area from their affected houses.The housing residence was first built and resided in 2013, and since then, it began to expand until now and has at least 84 hectares with 2331 housing units.The area's development rapidly occurred after relocating the 2018 Palu earthquake survivors.Another reason for choosing this location as a relocation site was that it only has a minor liquefaction effect in the form of sand boils compared to Petobo's flow liquefaction event [10,11].
This study will evaluate the influence of the soil behavior in the Kelapa Gading Residence, which previously experienced sand boils during the 2018 earthquake, on the area's vulnerability to liquefaction.The research location focused on the residence's already developed and future development.The affected areas of sand boils were manually recorded by interviewing site residents.The Cone Penetration Test was then conducted at 3 locations within the residence, including 1 location of hand boring, to collect basic knowledge about the soil (Figure 4).This data will also determine the area's soil behavior and liquefaction potential.

Site Conditions
Kelapa Gading Residence, Sigi Regency, Central Sulawesi, is situated on the Palu River valley's alluvial, flood, and old river channel deposit formation (Figure 5) (Qal) [10].These lower-energy fluvial environments (i.e., overbank flooding of a river) contain finer-textured alluvium, typically clays, silt, and fine sands.Furthermore, this area's sediment deposits also contain gravel, sand, mud, and coral limestone affected by the shallow marine environment.The presence of limestone mainly creates a sandy alluvium texture, allowing water to be stored in the voids between soil particles and become saturated [10,12,13].This will then increase the vulnerability of the soil in the area to liquefaction [14].Another critical factor of the Kelapa Gading Residence that makes it susceptible to liquefaction is the area located approximately 3 km from the Palukoro Fault.Palukoro is an active fault with a slip rate of less than 58mm/yr.It has an earthquake cycle of 130 years, except for the latest 7.5-moment magnitude earthquake of 2018, only after 109 years of stress energy accumulation from the fault movement [15].This indicates that the Palukoro fault has a rapid slip rate but a low seismicity level.This means the energy accumulation will be released into a major earthquake that likely causes liquefaction.

PGA determination
This study used PGA determination proposed by SNI 1726:2019, i.e., Indonesian standard, in determining seismic loading for designing buildings.The calculated PGA was published on https://rsa.ciptakarya.pu.go.id/2021/.The value of PGA on the CPT coordinates from the website was the median bedrock PGA that needs to be customized for the site near the fault with an adjustment of 1.8 times.Then, the surface PGA was calculated from this value by considering the site classification system by AASTHO (2012).Based on a study conducted by the Meteorological, Climatological, and Geophysical Agency of Palu Regional in 2019, Kelapa Gading Residence was in stiff soil (D) (Figure 6) with a PGA adjustment of 1.1 times.Table 1 shows the analysis results of the PGAM using this method.

Liquefaction potential based on grain characteristics using behavior type chart.
This study used a CPT Soil Behavior Type (SBT) chart, as Robertson (1990) proposed, to determine grain characteristics of the soil in the Kelapa Gading area [17].This chart was updated by Robertson in 2010 (Figure 7 a and b) with the revision of the non-normalized SBT index (ISBT) by directly using CPT measurements [18].The ISBT is defined as the equation below: The result will then determine the vulnerability of the soil to liquefaction with the sands (clean sands to silty sands) being the most vulnerable soil type [7].

Cone Penetration Test (CPT) Based Liquefaction Triggering Procedure
Boulanger and Idriss (2014) proposed the latest simplified CPT-b liquefaction triggering procedure to determine a liquefiable soil layer [19].This procedure was re-examined from the 2008 procedure with an update in magnitude scaling factor relationship, fines contents estimation, and most recent case histories to create a probabilistic version of the procedure [7,20].The liquefaction potential in safety factor (SF) form was obtained with the comparison of cyclic stress ratios (CSR) with the cyclic resistance ratios (CRR) of the soil (equation 2).If the SF value of the soil is ≥1.2 the liquefaction potential is zero [21].

𝑆𝐹 = 𝐶𝑅𝑅 𝐶𝑆𝑅
(2) Stiff fine-grained* The CSR value is alternatively derived from Seed and Idriss (1967) and elaborated by Idriss and Boulanger 2008 by considering the magnitude scaling factor (MSF) and overburden correction factor (K) which are expressed as equation ( 3) to (10) [7,19]: = 1 37.3 − 8.27( 1 ) 0.264 ≤ 0.3 (10) where v= vertical total stress at depth z, amax/g = maximum horizontal acceleration (g) at the ground surface (peak ground acceleration (PGA), and rd = shear stress reduction factor accounting for the dynamic soil profile response.Both MSF dan K values are considering their correlation with in-situ test results (CPT) [19].
To determine the CRR value, CPT data (qc) needs to be corrected for the overburden stress effect by using equations (11) to (13): = 1.338 − 0.249( 1 ) 0.264 (13) CN = overburden correction factor, Pa = atmospheric pressure, and qc1N is the resistance to penetration obtained in the same sediment at an overburden pressure of 1 atm if all other characteristics remain constant.Since all the values of the equation are interconnected, iteration is required to determine each of them.
The qc1N value then re-evaluated to find its equivalent adjustment for fines content to clean sand with equations ( 14) and (15): The final CRR value at the magnitude 7.5 and effective overburden stress at 1 atm can be calculated using the equation (16)

Liquefaction Potential Index (LI)
Liquefaction Potential Index (LI) was proposed by Iwasaki (1984) to evaluate the significant impact on structures above the soil caused by a degree of liquefaction at any given site [22].Sonmes (2003) modified the proposed classification to accommodate the non-liquefiable and the moderate liquefaction susceptibility, as shown in table 2 below [21].For the discretized CPT measurements less than 20m Luna and David (1998) calculate the liquefaction potential using the equation (17).

Results and Discussion
The soil behaviour type method was established by using CPT measurements that could determine the soil type and stratigraphy [18].This method produces a stratigraphy graph from the study area to define each layer's susceptibility to liquefaction.It is applied to the CPT measurements from three boreholes.The layer that contains sands (clean sands to silty sands) which range from 1.3 to 2.05 (ISBT), has the highest vulnerability to liquefaction.
Figure 8 shows the stratigraphy of the three boreholes in Kelapa Gading Residence with a depth of qc >250 at 5.8 meters.According to the ISBT analysis of the boreholes, the topsoil at boreholes S01 and S02 is a clay-organic soil type, which is consistent with the land use for plantation areas of rice fields and vegetables before conversion to habitation areas.The second layer is composed of clays (clays to silty clays) at S01 and S02.The third layer contains silt mixtures, and the fourth layer is mainly formed by sand mixtures (silty sands to sandy silt).The fifth layer contains clean sands to silty sands with a layer of dense sand to gravelly sands in the S02 borehole.The most susceptible layer is the layer that contains sands and sand mixtures.This analysis matches the sieve test (sieve number 200) conducted in a hand boring location with the fines content of the layer between -3.0 -(-4.0)meters was only 9.8%.Liquefaction analysis was conducted on each borehole up to 5.8 meters in depth (Figure 9).The magnitude used in this study was the latest and largest 7.5 Mw earthquake in Palu in 2018.For the PGA value, this study used the PGA of the same event near the Kelapa Gading Residence, recorded in Petobo (1 km from the residence) with the value of 0.334 g [10,23] and compared it to the estimated PGA value in the location with the value of 0.681, 0.685, and 0.688 g for future liquefaction potential.The PGA value was calculated using SNI 1726:2019 [24].Table 2 shows the analysis result at borehole S01.Based on the analysis result, a liquefied layer happened between the depth of 2.5 to 3.3 meters for the PGA=0.34g and 2.5 to 3.4 meters for the PGA=0.68 with the soil type of clean sands and sands mixture.The liquefied layer thickness was 0.8-0.9meters.The liquefaction potential index was then calculated on each borehole to estimate the severity level caused by the liquefied layers.Table 3 shows the liquefaction potential of each borehole with different PGA with severity levels low and moderate.The non-liquefied upper layers that have a thickness of 1.6 meters, which consist of 0.6 m clay-type soil (ISBT > 2.95) and 1 m silt mixture (2.6< ISBT <2.95) were able to hold the 0.8 m liquefied layer and prevented it to resurfaced during the 2018's Palu earthquake and discharged the excess pore water pressure only through sand boils.The findings of the study indicate that the behavior of the soil influences the susceptibility to liquefaction in the area.Specifically, it was observed that the presence of a clay-type soil layer as the topsoil decreases the vulnerability to liquefaction.The soil type in question possesses the ability to withstand the excessive pore water pressure originating from under the surface, so safeguarding the layer against liquefaction.This resistance can be attributed to the presence of a high proportion of fine-grained particles within the soil composition.In contrast, it can be observed that the sandy soil type in the region aligns with a liquefied layer, which is believed to have served as the origin of sand boils during the liquefaction event in 2018.The analytical findings indicate that a portion of the soil type exhibited vulnerability to liquefaction, whilst the deeper section shown resistance due to its higher density.

Conclusions
The liquefaction potential analysis has been conducted in the Kelapa Gading Residence based on the historical case of sand boils in the area during the 2018 Palu earthquake.The area is approximately 1 km from the liquefaction-affected site Petobo, which experienced flow liquefaction at the same disaster event.Geological feature of the area for the formation of the sediments: It is situated on the Palu River valley's alluvial, flood, and old river channel deposit formation (Qal) that mainly contains sandy alluvium texture, allowing water to be stored in the voids between soil particles and become saturated.On the other hand, the area is also located near the Palu-Koro fault (3 km), which makes it more susceptible to experiencing a higher-magnitude earthquake.Both geological features increase the area's vulnerability to liquefaction.
The soil investigation using soil behavior type methods was conducted.The stratigraphy result shows that soil layers in the approximate depth of 1.6-5.8meters contained a sands mixture and clean sands that were susceptible to liquefaction, while the CPT-based liquefaction triggering procedure proved that liquefied layer happened between the depth of 2.5 -3.3 meters with the thickness of 0.8 meters.On the other hand, the layer of clean sand from the soil type identification was not susceptible to liquefaction based on CPT-based liquefaction triggering methods due to its high-density level.The upper layer, which contains finer soil type (ISBT>2.6), which ranges from organic clay to silt mixture, prevented the liquefied layer from resurfacing and only induced sand boils.
To prevent future sand boils from happening in the residential area, the authors suggest that the area is planted with several locations of discharging wells until the minimum depth of 3.5 meters with the combination of a waterway on the green area of the site (park, or other open space) so that none of the housing would be affected by sand boils.This mitigation is considered adequate due to the already built area, and the soil dewatering was less possible because residents use the water from the wells for daily needs.

Figure 1 .
Figure 1.The central location of the Mw 7.5 earthquake event with the overall shakemap of all other quakes within the area [3].

Figure 4
Figure 4.The Kelapa Gading Residence area with CPT locations and sand boils recorded data, Open Sources Bing Map Aerial-Modified [9].

Figure 5 .
Figure 5. Geological map of Palu and surroundings.The study area comprises alluvial, flood, and river channel deposits [10].

Figure 6 .
Figure 6.Soil Classification of Sigi Regency Area 2019 by Meteorological, Climatological,and Geophysical Agency of Palu Regional-Modified[16]

Figure 7 .
Figure 7. (a) Non-normalized ISBT chart updated version, (b) table of soil classification based on soil behavior type index[18]

Table 1 .
PGAM Analysis Results for Each Boreholes.

Table 3 .
Liquefaction potential index for each borehole at Kelapa Gading Residence