Effect ofMoistureContent on the ShearBehaviour of aCompletely Decomposed Granite: An Experimental Study

To figure out the effect of moisture content on the shear behaviour of undisturbed completely decomposed granite (CDG) soil, a series of in situ and laboratory direct shear tests were carried out, under different normal stress and moisture contents. -e test results showed that the increasing moisture content could obviously weaken the shear strength, cohesion, and frictional angle. Two linear equations were proposed for estimating the cohesion and the frictional angle at any moisture content in the field. -e estimated strength parameters could be used for safety evaluation and/or engineering design of CDG cutting slope. -e rate of reduction of the cohesion was much higher than that of the tangent value of frictional angle. Due to the disturbance of laboratory sample and size effect, the strength parameters obtained from the in situ test were a little bit higher than those from the laboratory test.


Introduction
Rainfall-induced landslide usually causes catastrophic disaster, for example, a slope in Shenzhen, China, failed after a heavy rainstorm and killed 78 people in 2015 [1]. e main reason for this kind of failure is that the soil mechanical properties decreased with the increasing water contents resulting from rainfall infiltration. e mechanical behaviour is seriously influenced by the moisture content both for undisturbed soil (usually encountered in cutting slope) and re-compacted soil (filled slope) [2,3]. In situ tests can directly show the mechanical behaviour of the soil in construction location. However, because of inconvenience and high cost, in situ tests are seldom carried out. Instead, laboratory tests, i.e., direct shear test and triaxial test, are usually performed to obtain the mechanical parameters used for the safety evaluation [4][5][6]. However, due to the disturbance of soil specimen during sampling, transportation, and installation, the mechanical parameters obtained from the lab test are usually lower than those in the field, which would result in the rise of the inaccuracy of numerical simulation and the difficulty to do the safety evaluation [7,8].
Completely decomposed granite (CDG), as a main component of weathered granitic crust, is widely distributed in Malaysia, Brazil, South Korea, and especially Southeast China [9][10][11]. Due to the humid and warm climate in these areas, the thickness of CDG layer is usually in the range of 10 m-30 m. So, CDG is one of the most common soils encountered and is usually used as filled geomaterial. In Hong Kong, lots of artificial slopes were filled by re-compacted CDG soils in 1970s, and because of the lack effort of compaction during construction, the compaction degrees of these slopes were very low (less than 80%) [12,13]. When the loose soils encountered water, the shear strength decreased dramatically, which resulted in many failures during or soon after heavy rainstorm every year. e shear strength can be described by the Mohr-Coulomb criterion with two material parameters, namely, the cohesion c and the internal frictional angle φ as follows: where τ f is the shear strength and σ is the normal stress at the failure plane. To find out the influence of water content on the shear strength of loose compacted CDG, a series of laboratory direct shear tests were carried out, and the strength parameters c and φ were obtained by linear fitting, which could be then adopted for engineering design [14,15]. In Southeast China, undisturbed CDG was encountered in many cutting slopes. e mechanical behaviour of this undisturbed soil has been studied at laboratory in recent years [16][17][18], but seldom researched in the field. In this study, to investigate the influence of moisture content on the shear mechanical behaviour of undisturbed CDG, both in situ direct shear tests and laboratory tests were performed with various normal stresses and water contents. e strength parameters c and φ were obtained by linear fitting of test data.

Test Material and Procedure
e in situ direct shear tests of undisturbed CDG soil were conducted on a cutting slope located at Huizhou city, China, as illustrated in Figure 1. e soil was weathered from the Late Yanshanian granite according to the geological report. From visual inspection, the parent rock has been completely decomposed, but some of the original fabric and texture could be observed clearly. e soil could be easily crumbed by hand, and it was mainly brown and gray in color with some flesh red patches. According to the classification system proposed by Geotechnical Engineering Office [19], the soil could be classified as completely decomposed granite (Grade V).
Basic property tests were performed in accordance with BS1377 (British Standards Institution, 1990) [20], including specific gravity tests, Atterberg limit tests, sieving, and hydrometer analysis. e properties of the soil are shown in Table 1, and the particle size distribution curve is illustrated in Figure 2. e soil can be classified as sandy clay (CL) according to the Unified Soil Classification System (ASTM 2017) [21]. e in situ direct shear tests comprised of three series, for which the moisture contents were different. One series tests were carried out right after the soil block samples had been exposed to sunshine by half month, and so the moisture contents were lower than the natural one (Table 2, si1 to si6). e other series was performed the next day after a heavy rainstorm, as the samples reached moisture equilibrium, and the water contents were higher than the natural one ( Table 2, si7 to si11). e third series tests were performed immediately after the block samples were created, and the moisture contents were natural (Table 2, si12 to si16). e soil taken from shear band of each block sample was used to measure the moisture content when the shearing was finished. e procedure of the in situ tests is described as follows: (1) Preparation of block samples: three test pits were excavated at the construction site, for which the depths were about 1.5 m. In each pit, several block samples were formed by using pickaxe and saw to cut the ground soil to the sizes of 60 cm length × 50 cm width × 50 cm height, as shown in Figure 1. e samples in one pit were tested in dry condition and those in another pit were sheared in moist condition, and the soils in the third pit were tested in natural condition, which correspond to the three series in situ tests as mentioned previously.
(2) Installation of loading device: the reactive force system was then set up, which included several cuboid stone blocks, two shape steels, and two jacks, as shown in Figure 3. en, the shear box was assembled, which consisted of four steel plates and several bolts. e steel plates had the thickness of 20 mm to make sure enough stiffness so that the deformation of the box itself would be very small during the test. en, four dial indicators were installed to measure the displacements of block sample, two for vertical displacement and the others for horizontal displacement. One jack was then placed on the top of the sample for applying normal force, and another jack was installed on the side to apply shear force. In order to let the normal force could be applied onto the block sample uniformly, a steel plate was put between the upper jack and the sample. A part of the test system is shown in Figure 4. As shown in Figure 4, the block sample was higher than the shear box. It was because that the sample was very difficult to be cut in the field. e high soil sample would not change the normal stress on the shear surface, so it would not affect the results of the tests.
(3) Loading procedure: the predetermined normal force was then applied on the sample by the upper jack. For each vertical load application, the vertical deformation of the sample was treated to be steady if the increment was less than 0.1 mm in 5 minutes.
As the vertical displacement was stable, the shear force was applied by several grades until failure occurred. e shear force in the field test was not applied by constant rate of displacement, unlike laboratory test, and it was applied by constant force in each grade. In each grade, the horizontal and vertical displacements of the sample were measured. e shear force of each grade would not be applied until the displacements of upper grade remain steady. e normal force was controlled to be constant during shearing.
e samples for the laboratory test were obtained by pushing cutting ring into the ground soil at the bottom of pit, immediately after the in situ tests were finished, to make sure that the moisture contents of the ring samples were close to that of the in situ samples as possible. e size of ring sample was 6.18 cm diameter × 2 cm height. e moisture content of each ring sample was measured after the lab test. e details 2 Advances in Civil Engineering of in situ and laboratory tests are listed in Table 2. e in situ test in humid condition with the normal stress of 300 kPa was terminated at the beginning of shearing because of a mechanical problem, so there were only five completed field tests in humid condition.

In Situ Test
Results. e stress-strain curves of the undisturbed CDG under dry, humid, and natural conditions obtained from the in situ test are shown in Figures 5-7, respectively. e void ratio e of each sample listed in the figures was measured before the test.
It can be seen that all the block samples showed strainsoften behaviour during shearing, despite different moisture contents. At the initial stage, the shear stress increased with the increasing shear displacements. e soil structure afforded the main resistance, until the peak shear stress was reached. en, with the collapse of soil structure, the shear stress decreased rapidly with the shear displacement to a steady value. e residual shear stress was almost supplied by the friction which was dominated by the frictional angle of the soil and the normal stress (σ tan φ). In each condition, higher normal stress resulted in higher peak shear stress, except sample si4 (σ �150 kPa, Figure 5(a)), for which the peak stress would be expected to be higher than that of sample si3 (σ �100 kPa). It can be noted that the void ratio of si4 was 0.584, obviously higher than that of the other samples, which means the sample si4 had a lower density.
at would result in a lower peak shear stress than si3, although its normal stress is higher. In this case, the influence of density on the peak shear stress was greater than the normal stress. e vertical displacements of the soil samples are also shown in the figures. e upward displacement, which means dilatancy, is defined to be negative value. Under low normal stress, the soil showed strong dilatancy during shearing, and with the increase of the normal stress, the soil behaved to be contractive at the beginning and turned to be dilative at high strain. As expected, the soil tended to be contractive during all the shear process under high normal stress. In each moisture condition, the higher normal stress resulted in greater shear compression, except sample si4 which was thought to be a scatter. As mentioned earlier, the void ratio of si4 was higher than that of others, and the high e resulted in an unexpected compression ( Figure 5(b)). e curve of si4 was expected to lie between si3 and si5 if their void ratio were similar. However, the high void ratio of si4 meant that there existed more void which could be compressed during shearing than other samples. So, si4 showed a high compression, even that the normal stress was lower than si5. With the same normal stress, the vertical displacement at the end of shearing in the dry condition was always lower than that in the humid condition, showing more dilatancy. e increasing moisture content could increase the shear compression.   Advances in Civil Engineering e test data of the peak and residual shear stress versus the normal stress of each in situ sample are illustrated in Figure 8. e stress data of the samples in natural condition are not illustrated for clarity. Linear fitting is used to obtain the strength parameters, and the equations are also given in the figure. From the fitted equations, the strength parameters from the in situ test can be obtained, as shown in Table 3.
In the in situ test, the cohesion and friction angle obtained in the dry condition were 88 kPa and 44.2°, respectively. In the natural condition, for which the moisture content was a little higher than that of the dry condition, the cohesion and frictional angle were 70 kPa and 41.9°. As the moisture content increased, in the humid condition, the peak shear stress with the same normal stress was obviously lower than that in the dry and natural conditions (Figure 8), and the cohesion and frictional angle were 36 kPa and 35.4°. e strength parameters in different moisture contents are shown in Figure 9. It can be seen that both the cohesion and frictional angle decreased with the increasing moisture  contents linearly. Besides, the rates of reduction of the cohesion and tangent value of frictional angle were 59% and 27% as the moisture content changed from 4.1% (dry condition) to 20.4% (humid condition). e equations of the fitted lines for cohesion and frictional angle are listed as follows: where c ′ (kPa) and φ ′ (°) are the cohesion and the frictional angle at the moisture content of w. With the two equations, the cohesion and the frictional angle at any moisture content can be estimated for safety evaluation and/or engineering design of cutting slope of CDG soil.

Laboratory Test
Results. e stress-strain curves of the undisturbed CDG in the dry, humid, and natural moisture conditions obtained from the laboratory test are shown in Figures 10-12, respectively.
All the samples in the laboratory test showed the strainsoften behaviour, same as the block samples in the field test. In each humidity condition, the shear strength increased with normal stress, except sample la18 (σ �150 kPa), for which the peak shear stress was obviously higher than that of sample la19 (σ � 200 kPa) because of its very low void ratio (high density). With the similar void ratios, the peak stress in dry condition was higher than that in humid condition under the same normal stress, which means the increasing moisture content can weaken the shear strength of the undisturbed CDG. Both in the dry and natural conditions, all the samples showed negative vertical displacements (dilatancy) at the end of the test. In the humid condition, with the increasing normal stress, the samples turned from dilative to contractive. Sample la17 (σ �100 kPa) showed a little dilatancy during all the shear process, and the sample la19 (σ � 200 kPa) behaved to be contractive. It can be noted that the sample la18 (σ � 150 kPa) showed a strong dilatancy at the initial stage and trended to a negative value at the shear displacement of 7 mm. e displacement curve of la18 was expected to lie between la17 and la19. However, the low void Advances in Civil Engineering 5 ratio, which means a high density, of this sample would result in a strongly dilative behaviour during shearing, and this strong dilatancy would result in a dramatic peak shear stress, as shown in Figure 11. e test data of the peak and residual shear stress versus the normal stress of each laboratory test are illustrated in Figure 13, and the strength parameters obtained are shown in Table 4. e strength parameters obtained from the lab test in different moisture contents are shown in Figure 14.
e cohesion and frictional angle obtained from the laboratory tests were 85 kPa and 37.1°in the dry condition    and decreased to 30 kPa and 31.7°in the humid condition. e rates of reduction of the cohesion and tangent value of frictional angle were 65% and 18%. As seen in Figure 14, both the cohesion and the frictional angle decreased with the increasing moisture content linearly, same as the field tests.
By comparing the strength parameters obtained from the in situ tests and the laboratory tests, it can be concluded that the strength parameters obtained from the in situ test were slightly higher than those from the laboratory test with similar moisture content. is was due to the reason that the disturbance during sampling, transportation, and installation would weaken the soil structure of the laboratory samples, and the size effect could also affect the shear strength. In both the in situ and laboratory tests, the moisture content could obviously influence the shear behaviour of the undisturbed CDG soil. Besides, the strength parameters, cohesion, and frictional angle decreased with the increasing moisture contents linearly. e rate of reduction of the cohesion was much higher than that of the tangent value of frictional angle, which meant that the bonds of undisturbed CDG were much easily influenced by the moisture content than the fabric. Further research is required to find out the mesoscopic mechanism for this behaviour.

Conclusions
is paper presents the test data and the interpretations on the shear strength characteristics of undisturbed CDG soil in its unsaturated state. Direct shear tests in field and laboratory were carried out with different moisture contents and normal stress. e following conclusions are drawn based on the study presented in the previous sections: (1) e normal stress has a remarkable influence on the shear strength of undisturbed CDG. With similar void ratio, the higher normal stress would result in higher shear strength and more shear compression. However, sometimes a large difference in void ratio (density) could result in an unexpected shear behaviour. (2) Moisture content plays an important role on the shear strength characteristic of undisturbed CDG. e increasing moisture content could obviously weaken the shear strength, the cohesion, and the frictional angle. Two linear equations were proposed for estimating the cohesion and the frictional angle at any moisture content in the field. e estimated strength parameters could be used for safety evaluation and/or engineering design.
(3) With the increasing moisture content, the rate of reduction of the cohesion is much higher than that of the tangent value of frictional angle, indicating that the bonds in the undisturbed CDG are much easily influenced by the moisture content than the fabric.

Data Availability
e data used to support the findings of this study are included within the article.

Disclosure
is study was conducted as part of the employment of the authors.  Advances in Civil Engineering 9