日本地球惑星科学連合2022年大会

講演情報

[E] 口頭発表

セッション記号 H (地球人間圏科学) » H-DS 防災地球科学

[H-DS07] 地すべりおよび関連現象

2022年5月24日(火) 13:45 〜 15:15 201B (幕張メッセ国際会議場)

コンビーナ:千木良 雅弘(公益財団法人 深田地質研究所)、コンビーナ:王 功輝(京都大学防災研究所)、今泉 文寿(静岡大学農学部)、座長:今泉 文寿(静岡大学農学部)

14:00 〜 14:15

[HDS07-13] Pre-failure Kinematics Strongly Modulated by Shear Localization: An Experimental Study and Its Implication for the Failure-time Forecast of Landslide

*Chengrui Chang1Gonghui Wang2 (1.Divsion of Earth and Planetary Sciences, Kyoto University、2.Disaster Prevention Research Institute, Kyoto University)

キーワード:focasting landslide failure time, kinematics, shear localization

1. Introduction. Failure-time forecast is a critical and significant goal in landslide science regarding disaster prevention and mitigation. Landslides exhibit an extensive variety of kinematics as the materials respond to these inter-related variables and factors, which make the failure-time forecast a difficult task. There are, however, well-documented landslide cases, showing accelerating movement over timescales of weeks to decades. Such accelerating behavior universally arises in various forms of failure and natural phenomena, which are commonly perceived as precursory signals to forecast time to failure. A series of phenomenological models have been pioneered by several Japanese scientists to describe the accelerating movements of landslides or other materials. By simulating rainfall-triggered landslides in a large flume, Fukuzono (1985) found that surface velocity is proportional to the acceleration in a double-log plot in the vicinity of failure. This finding was generalized by the Voight model [d2Ω/dt2 = A(dΩ/dt)a] in which Ω denotes measurable quantities (e.g., displacement, tilting angle, acoustic emission count), and dot refers to time derivation (Voight, 1988). Two controlling parameters of α and A, defined by the accelerating trend of the quantities, are representative of kinematics prior to a major failure. However, this model does not take account of material behavior and failure mechanism, and the physical controls on the parameters (α and A) are unknown. Meanwhile, the picture of the connection between the Voight model and the landslide failure process is far from complete. We address these questions by simulating the accelerating creep of soils with novel ring-shear configurations and separating out the quantifiable effects on the values of α and A.
2. Methodogoly. Four types of soil samples were prepared including Silica sand No.7 (abbreviated as SS), and three mixtures consisting of mixing Silica sand No.7 and Bentonite. The total weight ratio of bentonite was varied as 10%, 20%, and 30%, and the mixture samples are termed as M1, M2, and M3, respectively. Ring-shear configuration (DPRI-5) was employed to simulate landslide initiation and activation that are triggered by loading or incision at the toe (increase in shear stress). Total normal stress and shear stress were applied to the sample to simulate the in-situ stress state of a given slope before the shear test. Then samples were brought to failure by increasing shear stress. To simulate reactivated landslide where shear failure occurred along pre-existing shear zone or sliding surface, shear tests were performed repeatedly, and the soil samples were re-consolidated with complete dissipation of pore-water pressure before the next test.
3. Results. The pre-failure kinematic features were analyzed both in the time domain and displacement domain, and the typical kinematic patterns were recognized and documented according to types of the time-velocity curve and the relationships between velocity and acceleration in a double-log plot.
For the first-time failures, velocity curves fluctuated in tests on SS samples and the kinematic pattern is described as “multi-accelerations”, and SS samples behaved the most significant volumetric variation before shear failure (Fig. 1). By contrast, both the variations in velocity curve and volume curve diminished gradually with the increase in clay content of the sample. Time-velocity curve was relatively smooth for clayey samples, which suggests a more accurate α value for the failure-time forecast. In addition, it was found that the pre-sheared sample frequently manifested a smooth time-velocity curve. The volumetric changes during shear are largely reduced for the pre-sheared samples irrespective of the types of material and a relatively unchanged α-value is favored (Fig. 2).
4. Summary. In brief, this study provides findings of a laboratory-based examination on a variety of soils by simulating landslide initiation and reactivation. Specifically, pre-failure kinematic behavior was observed and critical parameters of the Voight model were examined. The main conclusions could be summarized as follows. 1)Pre-failure kinematics was in association with the volumetric change in samples. The relatively invariant volume might indicate more localized shear deformation and fewer asperities. 2)The localized shear zone strongly regulated pre-failure kinematics, which might be attributed to the starting heterogeneity introduced by the pre-sheared history.