Experimental Investigation on Loading-Relaxation 3 Behaviors of Shear-Zone Soil

: This paper presents an experimental investigation of the loading-relaxation behaviors of reconstituted shear-zone soils 6 through drained triaxial tests. A multistage loading-relaxation approach is adopted to perform this test. The test aims to study two 7 main issues: the in ﬂ uence of relaxation on the mechanical response during reloading; and the in ﬂ uence of strain rates, loading increments, 8 and the relaxation time on the relaxation characteristics. The test results indicate that the stress-relaxation behavior of shear-zone soil is 9 dependent on the stress and strain levels. The loading patterns prior to the stress-relaxation process affect mainly the viscoelastic behavior 10 of the soil and subsequently in ﬂ uence the initial relaxation behavior. Moreover, a reloading after a stress-relaxation process gives rise to a 11 higher deviatoric stress owing to the viscoplastic hardening. It is further shown that the loading-relaxation behaviors can be interpreted by


15
Most reactivated slow-moving landslides in the Three Gorges shear zones (Miao et al. 2014;Wang et al. 2018b). The large de-18 formation within the shear zones usually contributes to the major-19 ity of the whole landslide motion owing to the viscous behavior of 20 the shear-zone soils. In general, the concept of viscous behaviors 21 consists of three aspects, namely, rate dependence, creep, and 22 stress relaxation (Feda 1992;Liingaard et al. 2004;Augustesen 23 et al. 2004). Rate dependence and creep deformation can be widely 24 observed in slow-moving landslides. It is reported that when af-25 fected by seasonal rainfall and periodic changes in water level, 26 most slow-moving landslides exhibit a creep deformation charac-27 terized by a stepwise trend (Wang et al. 2018b). The change in strain 28 rate in the shear zones could, in turn, affect the mobilized shear re-29 sistance of the shear-zone material and consequently influence the 30 creep deformation of the landslide (Hu et al. 2018). Owing to the sig-31 nificance of this property, a series of works have been carried out to 32 investigate the rate dependence (e.g., Scaringi et al. 2017;Hu et al. 33 2018) and creep properties (e.g., Li et al. 2017;Tan et al. 2018;34 Wang et al. 2018b34 Wang et al. , 2018c34 Wang et al. , 2020 of shear-zone soils taken from posure, e.g., excavation near the shear zone (Wang et al. 2018b) or 48 toe cutting owing to construction on the landslide area (Deng et al. 49 2016; Troncone et al. 2014). In these cases, a complex stress state 50 is imposed on the shear-zone material and loading-relaxation cou-51 pling is likely to occur. There are several reports on the loading-52 relaxation coupling effect for reconstituted clay in odometer tests 53 (Bagheri et al. 2019) and triaxial tests (Wang et al. 2017). In these 54 tests, the loading-relaxation coupling effects significantly influ-55 ence the viscous behavior of the tested materials. For example, 56 the rate and magnitude of the relaxed stresses increase with the in-57 crease in the prerelaxation strain and stress levels, and strain rates 58 (Bagheri et al. 2019). Likewise, a stress-relaxation process is 59 bound to affect the subsequent reloading behavior in a multistage 60 loading-relaxation process. However, there is a lack of relevant 61 research on shear-zone soils. Therefore, how the stress relaxation  were detected in the landslide (Wang et al. , 2016(Wang et al. , 2018a.

84
The large size and complex structure of Huangtupo landslide make 85 it an ideal study object to better understand the mechanisms of sim-86 ilar slides in this area. In 2012, the Badong experimental station, in-87 cluding an intensive monitoring network and an investigation tunnel 88 with five adits, was established by the Three Gorges Research Centre 89 for Geohazards. The adits exposed several shear zones of this land-90 slide for scientific experiments, such as in situ triaxial creep and di-91 rect shear tests (Wang et al. 2020;Tan et al. 2018).

92
It is noteworthy that the shear zones in the investigation tunnel 93 will be exposed to the air for a very long time owing to the time-94 consuming in situ mechanical tests. Without a lining support, stress 95 redistribution will take place in the soil, and the exposed shear zone 96 will move into the tunnel, leading to considerable stress relaxation 97 in the shear zone. Moreover, the stress-relaxation phenomenon was 98 also observed in in situ direct shear test. As shown in Fig. 1(d), for 99 example, a significant stress relaxation occurred at the end of the 100 in situ direct shear test when the shear displacement ceased. This 101 large amount of stress relaxation was ascribed to the accumulation 102 of structural defeat over time in the soil under the sustained strain.

103
The stress was anticipated to be decreasing with time. However, 104 further relaxation was not measured in the in situ test. According

105
to the rheology theory of stress relaxation, the stress may decrease

126
The material in the shear zone is characterized as gravelly soil 127 with inhomogeneously embedded coarse particles. It is reported 128 that the inhomogeneity of the shear-zone soils has a significant in-

129
fluence on their viscous behaviors (Wen and Jiang 2017). More-130 over, previous experiments with intact shear-zone soil of this 131 landslide usually give rise to very scattered results (Wang et al. 132 2018b;Li et al. 2017;Tan et al. 2018;Wang et al. 2020). To high-

133
light the loading-relaxation coupling effects, therefore, remoulded 134 soil with a particle size less than 5 mm was used in this study. Cylin-135 drical specimens of 50 mm diameter were prepared with equal 136 heights of 100 mm for the tests. The ratio of the specimen diameter 137 to the maximum particle size is 10, which satisfies the requirement 138 for minimizing the boundary effect. The remoulded soil was mixed 139 with water to reach a natural water content and, then, the material 140 was stored in a covered container for 2 days. To achieve better homo-141 geneity, the specimens were compacted into six layers using a split 142 mould with a circular cross-section to achieve the desired bulk den- i.e., UR test), and combined loading-relaxation tests (LR tests).

152
The program for all involved tests is given in  sures of 400, 1,000, and 1,300 kPa were performed for comparison.

166
The LR tests were carried to investigate the loading-relaxation ef- can be evaluated using a single specimen in the LR test (Sheahan 172 et al. 1994). An axial strain-controlled approach was employed dur-

173
ing the relaxation tests, in which the axial strain was kept constant 174 when the desired strain was reached, but the radial strains were 175 free to evolve during the testing process. This boundary condition 176 is consistent with that of the soil in the shear zone.

276
The volumetric strain of the LR1 test is greater than that of the 277 LR2 test with a variable rate of strain, as shown in Fig. 7(b).
278 Fig. 8   indicates that the strain rate at the previous loading stage has a re-291 markable influence on the relaxation response. This effect, however, 292 gradually disappears after the transition point.

293
Effect of Loading Increment 294 Fig. 9 shows the comparison of the LR1 and LR3 tests. The differ-295 ence in loading increments gives rise to different mechanical re-296 sponses. For example, the LR3 test yields a larger deviatoric stress 297 at the axial strain ranging from ε a = 2−9%. The difference in devia-298 toric stress, however, is compensated at ε a = 9% after the LR1 test 299 experiences the same number of relaxation processes as that in the 300 LR3 test. Thereafter, the LR1 and LR3 tests gain almost the same de-301 viatoric stress, until the LR3 test undergoes one more stress relaxa-302 tion at ε a = 17%. Although the LR3 test experienced a larger 303 volume change during the entire loading-relaxation process, as 304 shown in Fig. 9(b), it is believed that the stress curve of the LR3 305 test will coincide with that of the LR1 test after ε a = 18%. In addi-306 tion, the LR3 test also shows a switch of strain hardening to strain 307 softening in the stress-strain curve, whereas this transition point ap-308 pears at ε a = 12% rather than ε a = 9% in the LR1 test.

309
The comparison of stress relaxation responses from the LR1 and 310 LR2 tests is presented in Fig. 10. A similar stress-relaxation behavior 311 is observed in these two tests. Likewise, the transition point appears at 312 approximately 3 min, earlier than that observed in the LR1 test, after 313 the relaxation test starts. Compared with the LR1 test, the LR3 test 314 experienced one more relaxation process at the earlier stage of axial 315 strain. This additional process gives rise to a strengthening effect to 316 the soil. As a result, at the same strain level, the stress-relaxation ratios 317 in the LR3 test are larger than those measured in the LR1 test.

319
Previous investigation has shown that neither test LR2 nor test 320 LR3 exhibited an obvious third stage of stress relaxation. This 321 is because the stress was relaxed for only 1 day in both tests.

322
Thereafter, in the LR4 test, the duration of the stress-relaxation 323 stage was extended to 3 days, whereas the other conditions 324 were kept the same as in the LR1 test.

325
The coupling effects of relaxation time on the mechanical behav-326 ior of shear-zone soil during reloading are shown in Fig. 11. In com-327 parison with the LR1 test with 1-day relaxation duration, the time 328 effect is more pronounced in the LR4 test with 3-day relaxation.

329
As shown in Fig. 11(a), the reloading gains much higher deviatoric 330 stress under the strain level of 6%. For the subsequent reloading at 331 higher strain levels, the difference due to the longer previous relaxa-332 tion duration becomes more visible. It is interesting to note that the 333 deviatoric stress q 1 is almost equal to q 2 , with q 1 and q 2 being the 334 deviatoric stress at the beginning of the third reloading state in 335 the LR1 test, and at the end of the first reloading state in the LR4 336 test, respectively. Fig. 11(b) presents the volume change of the LR1 337 and LR2 tests. Clearly, the LR4 test gained much greater volume con-338 traction than that in the LR1 test during the entire relaxation duration.

339
Nevertheless, for the loading process, the LR1 test obtained a slightly 340 larger volume contraction, as shown by the dashed line in Fig. 11(b).

368
Although the axial strain remains unchanged, the volumetric strain 369 still increases owing to the radial strain. Therefore, the radial strain 370 presents the whole volume change, and it is obviously larger than 371 that caused in the preliminary stage.

372
The viscoplastic hardening can be clearly observed in the re-
where dɛ rev and dɛ vp denote the viscoelastic and the viscoplastic loading. The stress states are obtained from the LR1 test with ε a = 3%−6%.
The initial relaxation velocity v ri , reflecting the activity of inter-454 nal stress reduction and transition (Wang et al. 2017), is defined as v ri = Δq r t r (4)

455
where Δq r is the change in the deviatoric stress before the transition 456 time t r . For the sake of simplicity, the transition time t r = 60 s is 457 adopted in this work.
458 Fig. 15(a)  fluence the reloading behavior through relaxation time.

483
In this paper, we report on the loading-relaxation coupling effect of processes. In addition, the behavior upon reloading after re-494 laxation is initially in an viscoelastic regime, which can be 495 enhanced by increasing the relaxation time.

496
(2) The preceding loading patterns affect mainly the viscoelastic 497 behavior of the soil, and subsequently influence the initial re-498 laxation behavior. The loading increments and relaxation du-499 ration influence the subsequent relaxation behavior through 500 the viscoplastic hardening effects. In terms of the relaxation 501 duration, we found that longer previous relaxation leads to 502 observable differences in the initial drop of the stress as 503 well as the relative stress reduction.

504
(3) The loading-relaxation coupling can be interpreted with elas-505 toviscoplastic theory, provided the viscoplastic strain con- to thank the Otto Pregl Foundation for financial support in Austria.