Influence of Intra-Particle Saturation Ratio on Strength Degradation of Pumice Soil

Abstract

The 2018 Hokkaido Eastern Iburi Earthquake occurred on 6 September 2018 and caused more than 7000 landslides in and around Atsuma and Abira towns. The geology of this area is mainly formed by pyroclastic fall deposits. The rain gauges in the epicentral area showed little rain from typhoon Jebi (5 September 2018) and heavy rain in the middle of August 2018, about 3 weeks before the earthquake. A series of direct shear tests and single particle fragmentation tests were conducted to investigate the effect of soaking time on the strength reduction of Tarumae d (Ta-d) pumice. The median Single particle fragmentation strength of Ta-d particles decreases with the soaking time. The Intra-particle saturation ratio of Ta-d has increased with the logarithmic value of the soaking time. Samples were soaked in water for 1, 6, 24, 168 and 508 h before shearing. Peak shear strength was decreased with the soaking time. The residual strength is about 2 kPa for each soaking condition, which shows that there is no significant variation in residual strength with soaking time. The friction angle shows a sharp reduction from the peak value during the first 800 mm of shear strain and thereafter gradually stabilizes to the residual value. The peak shear stress of Ta-d pumice is reduced with the intra-particle saturation ratio. The mobilized frictional coefficient of Ta-d is not sensitive to the intra-particle saturation ratio.

1 Introduction

The 2018 Hokkaido Eastern Iburi Earthquake struck the eastern Iburi region of Hokkaido, the northern Japanese island, on 6 September 2018. The earthquake had a moment magnitude (Mw) of 6.6 corresponding to a JMA (Japan Meteorological Agency) magnitude (M_j) of 6.7 and a maximum seismic intensity of 7 on the JMA 7-stage seismic intensity scale. This earthquake exerted a severe socioeconomic impact on Hokkaido and all of Japan. (Osanai et al. 2019). This earthquake also caused a large number of shallow landslides and several large-scale deep-seated landslides.

The soil over the Atsuma area is mainly formed by the pyroclastic fall deposits caused by the three main volcanos. Specifically, the Tarumae volcano is abbreviated as “Ta” The eruption age of the typical deposits of Ta-d is 9000 years (Geology of the Chitose District 1980; Furukawa and Nakagawa 2010). Earthquake-induced slope failures in volcanic areas are of major concern due to the presence of problematic volcanic soils (Chiaro et al. 2018) and Rainfall and water content are the most influential factors for the ground stability when it is subjected to earthquakes (Kiyota et al. 2017).

During the investigation after the landslides (2018-09-18), Zhou et al. 2021 found that the volcanic deposits on the mountain slopes were close to saturation. Groundwater gushed from some landslide surfaces suggesting that heavy rainfall may have significantly contributed to the landslides during the earthquake. Typhoon Jebi hit the Hokkaido region on fifth September 2018, 1 day before the earthquake. However, rain gauges in the epicentral area showed little rain since the heavy rain in the middle of August 2018, about 3 weeks before the earthquake. Figure 1 shows the rainfall patterns recorded by the Automated Meteorological Data Acquisition System (AMeDAS).

Fig. 1

Rainfall received by the Atsuma area in July, August, and September 2018—JMA Data stations of Atsuma, Abira and Mukawa

Pumice particles are highly crushable, compressible, and lightweight due to their vesicular nature (Chaney et al. 2001; Kikkawa et al. 2013; Pender et al. 2006). They can retain water in their intra-particle voids for a while. Based on the on-site field reconnaissance from 2018-09-10 and the preliminary report (Hirose et al. 2018), the sliding zone of the majority of the Atsuma landslides is located in the Ta-d layer. Thus, the pumice layers are the most causative of the multiple landslides (Kawamura et al. 2019). In this study, a series of direct shear tests and single particle crushing tests on the pumice samples were conducted to clarify the strength reduction mechanism of Ta-d pumice soil with the soaking time. This paper discusses the effect of soaking time on the mechanical properties of the collapsed Ta-d pyroclastic fall deposits distributed over this area.

2 Material and Method

2.1 Material

Disturbed samples from Ta-d pumice soil were collected from Atsuma landslide sites. Basic soil parameters are listed in Table 1.

Table 1 Physical properties of Ta-d pumice soil

A series of physical tests were conducted: soil in-situ density tests, particle size distribution tests, Moisture content tests, and Specific gravity tests. Each test was implemented based on the JGS standards JGS 0191, JGS 0131, JGS 0121, and JGS 0111, respectively.

The results are summarized in Fig. 2 and Table 1. It is evident from the grain size distribution that the soil is mostly composed of coarse-grained soil particles and the mean particle diameter (D50) is 5.4 mm. In-situ density is 1.05 g/cm³ with water content 208% at the sampling location and dry density was set to 0.34 g/cm³ for direct shear test.

Fig. 2

Particle size distribution of sampled Ta-d soil

2.2 Test Method

2.2.1 Single Particle Fragmentation Test

Single particle fragmentation tests were performed by selected Ta-d particles whose particle diameter equals D50. The Samples were sieved from 6.7 mm and 4.75 mm sieves, and particles retained in 4.75 mm sieves were used for the test. Six batches of soil particles were prepared, and five were soaked in distilled water for 1, 6, 24, 168 and 508 h to make different soaking conditions. Distilled water is used to eliminate any undesirable impacts of chemical compounds. Another batch was used to evaluate single-particle strength in a dry state. Soil particles were placed between two smooth platens of the single-particle crushing apparatus with a maximum loading capacity of 5kN. The load was monotonically increased at a constant displacement rate of 0.1 mm/min until the particle was broken or crushed. More than 20 soil particles were tested in each batch to obtain statistically meaningful averages of the individual measurements on a single particle to account for the size variability and other heterogeneity of individual soil particles. (e.g., shape or geometry, surface roughness) (Liu et al. 2021). Hence, more than 120 particles were tested in six batches. Particles were selected with higher angularity (>0.6) and sphericity (>0.6). because particles with lower sphericity and angularity showed irregular failures, which does not correlate with particle size, as observed by (Nakata et al. 2001) Particle fragmentation strength was calculated by using Eq. 1.

$$ \sigma f=\frac{0.9 Nf}{d2} $$

(1)

Where σf is the particle fragmentation strength, Nf is the load at failure, and d is the distance between the two points of contact between the particle and the platen. d should be the smallest dimension of the irregular particle. (Hiramatsu and Oka 1966; Jaeger 1967). Equation (1) includes a correction coefficient of 0.9 to account for the effects of imperfect particle geometry and surface roughness (Liu et al. 2021).

2.2.2 Determination of Intra-Particle Saturation Ratio

Pumice is characterized by the vesicular nature of its particles; each particle contains a dense network of fine pores, some of which may be interconnected and open to the surface, while others may be completely isolated within the particle. They are therefore easily absorbent, and a considerable amount of water could be trapped in the pores. The amount of saturation of pores inside the particles will be referred to as the Intra-particle saturation ratio hereafter. To identify the change in the intra-particle saturation ratio with soaking time, we conducted a series of grain soaking tests for different particle sizes as 6.7 mm, 4.75 mm and 2.8 mm.

Five batches of particles were prepared for each size, and one batch consisted of 10 oven-dried particles. Initially, selected particles were scanned using a 3D scanner, and the volume of the particles (V_T) was measured using the scanned 3D model. The weight of each batch was measured in dry condition and subsequently soaked in distilled water. The weight of the particle batches was measured at the defined soaking time as 1, 6, 24 and 168 h, as same as the single particle fragmentation strength test. These data calculated the intra-particle saturation ratio (S) for each particle size as follows.

$$ \mathrm{S}=\mathrm{Vw}/\mathrm{Vv} $$

(2)

where V_V is the volume of voids inside the particle (volume of intra-particle voids) and V_T &V_S are the total volume and volume of soils respectively. V_W is the volume of water absorbed by the particle.

$$ {\mathrm{V}}_{\mathrm{V}}={\mathrm{V}}_{\mathrm{T}}-{\mathrm{V}}_{\mathrm{S}} $$

(3)

V_V is the volume of intra-particle voids. V_S was calculated in terms of specific gravity. V_W was calculated by using the measured weight of the particle at each soaking time.

2.2.3 Direct Shear Tests

We conducted a series of cyclic direct box shear tests under constant-volume conditions for reconstitute samples to investigate the shear strength property of pumice fall deposits. Particles larger than 4.75 mm were removed, and the rest part of an in-situ particle size distribution curve was maintained in each test. A schematic view of the modified direct shear apparatus used in this study is shown in Fig. 3.

Fig. 3

Schematic diagram of modified direct-shear test apparatus (modified based on Kiyota et al. 2011)

The initial dimension of the specimen was 200 mm in length, 200 mm in width, and 108 mm in height. The apparatus has the following essential features: (1) possible feedback control on both normal load and shear load to impose any prescribed stress path in the shear stress-normal stress space; (2) a lower shear box moving on a very low-friction rail, with two friction load cells to evaluate any friction at the bottom of the lower shear box; (3) Monotonic and cyclic loading test can conduct in constant volume and constant stress conditions. (Sharma et al. 2017; Kiyota et al. 2011).

The effect of the opening size between two halves of the direct shear apparatus was investigated. As per the results, it was decided that 10 mm is a suitable opening size for the Ta-d pumice soil. Flexible sponges were attached to the lower shear boxes to fill 10-mm openings and prevent particle leakage during experiments. These sponges were flexible and negligibly affected the vertical and shear stresses (Sawatsubashi et al. 2021).

Disturbed samples were first sieved and separated into different particle sizes. In preparation, sieved samples were mixed propositionally as per the in-situ particle gradation curve. In each test, the same initial particle gradation curve was maintained so that it is possible to compare the particle crushability after each shearing. The direct shear test sample was divided into seven layers. Weight and height per layer were decided as per the measured in-situ density. The sample was prepared layer by layer in pre-determined density using the dry tamping method. This method was used to obtain equal density throughout the sample. The confining pressure for each soil type is decided per the depth of soil layers in Atsuma slopes. After initial preparation, samples were consolidated under the prescribed confining pressure. Consolidated samples were soaked in distilled water for 1, 6, 24 and 168 h within the direct shear apparatus. For this purpose, an additional water-sealed box was attached to direct the shear apparatus from outside the shear box. The initial density of the direct shear sample was controlled to obtain equal density at the start of shearing in each soaking condition. The samples were soaked for a predetermined time, and subsequently shearing was done.

We conducted constant volume cyclic loading direct shear tests to identify the effect of soaking time on peak shear stress and residual stress. In the monotonic loading condition, the maximum shear displacement is 20 mm, and Ta-d samples did not reach the residual stress conditions at 20 mm shear displacement. Hence, strain-controlled cyclic loading tests were conducted for 100 cycles until the sample achieved the residual stress condition. In the cyclic shear process, the initial vertical stress was 30 kPa, and horizontal displacement was automatically controlled from zero to 20 mm and − 20 mm. After each sharing test, sieve analysis was conducted to measure the particle breakage.

3 Results and Discussion

3.1 Single Particle Fragmentation Test

As per the results, the single particle fragmentation strength (σ_f) of Ta-d pumice reduces with soaking time, as shown in Fig. 4. Analysis based on simplified quartiles shows that, on average, the σf remains the same until 168 h. Still, the strength at 504 h decreased by 48% compared to dry conditions. In addition, when comparing the first and third quartiles, there is no decrease with soaking time up to 24 hours, but there is a decreasing trend after 168 h. The variation of crushing strength of particles gets smaller with the soaking time. As discussed by McDowell and Bolton, the crushing strength of crushable soil is more variable than that of non-crushable soil because the individual particles of crushable soil often consist of several minerals. Even if they consist of a single mineral, the degree of the weathering is different (McDowell and Bolton 1998) In addition, the asperities on a particle with an irregular shape can break more easily. Therefore, crushing phenomena for crushable soils are expected to occur at relatively low stress levels.

Fig. 4

Change of single particle fragmentation strength with soaking time

3.2 Intra-particle Saturation Ratio

The intra-particle saturation ratio (S) increases with the soaking time, as shown in Fig. 5. The figure shows the average value for the S for a particular particle size. Smaller particles show a higher saturation ratio than the bigger particles at a given time. This trend is because smaller particles have comparatively smaller volumes of voids and could be saturated within a shorter period. The average value of the S shows a good correlation with the logarithmic value of the soaking time, as shown in Fig. 5.

Fig. 5

Change of Intra-particle saturation ratio of different particles sizes with log value of time (H)

3.3 Direct Shear Test

Constant volume direct shear test results show that the peak undrained shear strength (τ_peak) of Ta-d pumice is reduced with the soaking time. Figure 6 shows a gradual decrease in τ_peak as the soaking time increases.

Fig. 6

Stress-strain curves and stress paths of Ta-d in different soaking conditions

The τ_peak was reduced to 25 kPa at the 3 weeks (504 H) soaking from 40 kPa in dry condition. A similar trend was observed in the single particle fragmentation tests as well. In Fig. 6, the stress path shows a drastic reduction of shear stress within the first few cycles of the cyclic loading, and the value stabilizes at around 2 kPa by the 15th cycle. The stable value could be considered as the residual strength of the Ta-d pumice soil under undrained conditions. There is no significant variation in residual strength with the soaking time, as shown in Fig. 7. Allowable vertical deformation in content volume direct shear test is 0.05% as per JGS 0560–2020 (Japanese Geotechnical Society Standard (JGS 0560-2020) 2015). During the experimental program, it could not maintain the vertical displacement within the allowable limit, which may affect the peak shear stress value. This will be addressed in future studies.

Fig. 7

Peak and residual stress variation with soaking time

When the soaking time increases, the stress paths move away from the τ axis. In other words, the value of σ_v did not reach 0; the minimum value is increasing with soaking time increase. When the soaking time increases, the particle's crushability increases, and the volume change behavior decreases.

The τ_peak reduction in direct shear shows a good correlation with the S of Ta-d pumice. As shown in Fig. 8, peak strength reduces when the S increases. This strength reduction is observed in the single particle fragmentation strength test in the same manner, while S has no significant impact on residual strength.

Fig. 8

Peak and residual stress variation with Intra-particle saturation ratio (S)

The reduction of τ_peak may be caused by the particle strength reduction due to soaking. As shown in Fig. 6, σ_f reduced with the soaking time increased. The relationship between the peak shear stress and σ_f is shown in Fig. 9. Accordingly, there is very good evidence that the reduction of σ_f has caused the reduction in the τ_peak in the direct shear of the Ta-d pumice sample.

Fig. 9

Relationship of median particle fragmentation strength with peak and residual shear stress

When the sample is subjected to cyclic loading, the friction angle (∅) is changed accordingly. Figure 10 shows the variation of the mobilized frictional coefficient (μ_m) with cumulative shear displacement (Assumed C = 0). As per the results, the peak frictional coefficient (μ_p) in dry conditions is 1.3 and μ_p for 1H and 6H soaking conditions is about 1. Thereafter, the μ_m shows a sharp reduction by the tenth cycle and subsequently, no significant reduction until the 100th cycle. The residual frictional coefficient (μ_r) is 4.7⁰ for dry conditions and 3⁰ for 1H and 6H soaking conditions.

Fig. 10

Change of mobilized frictional coefficient with cumulative shear displacement

Konagai and Nakata 2019 discuss the uniformity of the mobilized friction angles of about 30 studied landslides in the Atsuma area. They measured the dimensions of 30 landslide masses in Atsuma, Hokkaido, as shown in Fig. 11.

Fig. 11

—Measured dimensions of Landslides. (Konagai and Nakata 2019)

A landslide mass with its initial length 𝐿₁ and cross-sectional area 𝐴₁ is assumed to have decelerated as it travelled over flat land and stopped completely with its final length 𝐿₂ and cross-sectional area 𝐴₂ immediately when the whole mass left the slope 𝐿₁. The variations of 𝐴₁ and 𝐴₂ along the direction of the dip (𝑥) are assumed to be substantially small and fluctuate little around their average values $\overline{A}$ ₁ and $\overline{A}$ ₂. An equation incorporating mobilised frictional coefficients was developed by equating the initial potential energy of landslide mass with the work done by the landslide during the sliding process.

Multiple linear regression analysis for the relationship among the measured dimensions of 30 landslide masses has given the average values of mobilized frictional coefficients of 0.165 on the slip surfaces. However, Konagai & Nakata concluded that the value of the mobilized frictional coefficient for a smaller and gentler slope, which might have been wetter than the others, could have been even smaller than this average value and was obtained to be 0.05. (Konagai et al. 2018, 2021; Konagai and Nakata 2019). The mobilized fictional coefficient value obtained by the experimental program is 0.052 and 0.05 for the 1H and 6H soaking conditions. Two values obtained from statistical analysis and experimental program are comparable.

4 Conclusions

To investigate the effect of the intra-particle saturation ratio on the strength characteristics of Ta-d pumice soil, we conducted a series of single particle fragmentation tests and direct shear tests. Based on the study, the following conclusions are drawn,

Ta-d Pumice is characterized by a low specific gravity (1.7 g/cm³) compared to non-volcanic soils. The Ta-d pumice samples had dry bulk densities of 0.34 g/cm³, lower than typical non-volcanic soils.

The intra-particle saturation ratio of Ta-d pumice correlates well with the Logarithmic value of soaking time.

The Median particle fragmentation strength of Ta-d pumice reduces by 21% with the increase of the intra-particle saturation ratio.

Peak shear strength in the direct shear test is reduced with the increase of the soaking time or intra-particle saturation ratio, and the residual strength is not sensitive to the intra-particle saturation ratio.

The friction angles in the direct shear test change with the cyclic loading. The peak friction angle is 53⁰ for dry conditions and 45⁰ for 1H and 6H soaking conditions. The residual friction angle is 4.7⁰ for dry conditions and 3⁰ for 1H and 6H soaking conditions. The mobilized frictional coefficient is about 0.05 for the 1H & 6H soaking conditions.