The soils used in this research are loess and filling in ground fissures in Jingyang, China. Blocks of samples were collected by the cutting method after stepped exploratory trenches had been excavated. Figure 1 presents the cross section of geology, the distribution of fissures and the sampling location. Two blocks of filling samples were collected near f12 (fissure 12) in depth of 12.5 m and two blocks of loess samples were collected at f12 in depth of 10.5 m. The loess is Q3 loess (Malan loess) and the color is yellow, whereas the filling is non-uniform containing stones and snail shells and the color is brown, darker than that of loess. Fracture filling about 5-7 mm in width exists in the sample of loess as illustrated in Fig. 2(a). It should be noted that the filling in loess is different from the collected filling sample as shown in Fig. 2(b).

Fig.1

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.1 Sample location, geological profile and distribution of fissures

Fig.2

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.2 Collected samples. (a) Loess; (b) filling

Natural loess specimens and natural filling specimens with 61.8 mm in diameter and 20 mm in height were prepared for the oedometer test; natural loess samples and natural filling samples with 39.1 mm in diameter and 80mm in height were prepared for the triaxial test. The samples were saturated by the vacuum saturation method before tests. The saturated remoulded oedometer samples and saturated remoulded triaxial samples were prepared by the following step:

(1) Air-dry the natural materials which were cut off when preparing natural oedometer samples and triaxial samples.

(2) Crush the air-dried natural materials and sieve materials to remove materials larger than 0.25 mm and measure the water content of sieved materials w₀.

(3) Measure the weight of sieved materials m₀; pave materials in disk; add proper amount of water by nebulizer to the materials and churn up the materials uniformly, the total water mass added can be calculated by the formula: mw=(w-w0)1+w0m0, w=15% in this study, which is the same as natural filling and near to the natural water content of natural loess; place materials in a plastic bag and put the bag in a moist chamber for at least two days to ensure the water content uniform in soil.

(4) Compact the materials in the sample compacting cylinder by two layers for the oedometer sample and obtain the oedometer sample with 10 cm in diameter and 3 cm in height. Compact the materials in sample compacting cylinder by four layers for triaxial sample and obtain the triaxial sample with about 39.1 mm in diameter and about 80 mm in height. The materials mass of each layer can be calculated by the formula: m=Gs(1+w)1+eVsn, where e is 0.80 for remoulded filling and 1.08 for remoulded loess, Vs is 235.6 cm³ for oedometer sample or 96 cm³ triaxial sample, n is the total layers.

(5) Extrude the oedometer sample from mould, use the ring to cut the sample and obtain the final oedometer sample with 61.8 mm in diameter and 20 mm in height. Triaxial specimen extruded from the mould is the final triaxial sample.

(6) Saturate the samples by the vacuum saturation method.

The physical properties of loess and filling are given in Table 1. As shown in Fig. 3, the loess is clayey loess according to the loess types defined by plastic limit and liquid limit [ 28] and the same result can be obtained according to the loess types defined by the grain size distribution curve [ 28], as shown in Fig. 4.

Tab.1

Tab.1

Tab.1 Physical properties of loess and filling property loess filling natural moisture content, % 16.9 15.0 natural density, g/cm3 1.52 1.73 void ratio 1.07 0.80 grain size distribution sand>0.05mm, % 12.1 9.5 silt 0.05~0.005mm, % 83.1 84.1 clay<0.005mm, % 4.8 6.4 atterberge limits wl, % 33 31.3 wp, % 18.9 14.6 Ip 14.1 16.7

Tab.1 Physical properties of loess and filling

Fig.3

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.3 Plasticity properties of loess and filling as compared with loess types defined by [ 27]

Fig.4

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.4 Grain size distribution curves of loess and filling as compared with loess types defined by [ 27]

The void ratios of filling samples vary between 0.65 and 0.82 due to non-uniformity and the void ratios of most triaxial samples are near 0.8. As the content of clay component in filling is higher than that in loess, the plasticity index of filling is therefore higher than that of loess.

Figure 5 shows the scanning electronic microscope (SEM) photos of the tested soils. In the photos, we can clearly see particles of natural loess and remoulded loess, aggregates of natural filling and remoulded filling. Loess is looser than filling on the whole.

Fig.5

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.5 SEM photos of loess and filling on vertical plane of sample. (a) Natural loess; (b) natural filling; (c) remoulded loess; (d) remoulded filling

Conventional oedometer cell and GDS triaxial apparatus were used to carry out oedometer tests and triaxial stress path tests respectively. The schematic diagram of GDS triaxial apparatus is shown in Fig. 6. The triaxial cell is Bishop & Wesley (1975) cell [ 29]. The working principle of GDS can be found in the Ref [ 30]. The stress path that GDS triaxial apparatus carries out can be a linear combination of radial stress and axial stress.

Fig.6

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.6 Schematic diagram of GDS triaxial apparatus

Table 2 shows the program of oedometer tests. The applied vertical pressure adjusted in time according to the obtained e-log p curves, that is, decreasing the increment of vertical pressure around turning section of e-log p curves. The vertical pressures shown in Table 2 are real values in the tests.

Tab.2

Tab.2

Tab.2 Program of oedometer tests samples load grades/kPa saturated natural loess 15, 30, 50, 65, 80, 95, 150, 200, 250, 300, 400, 600, 800, 1600,3200 saturated natural filling 15, 30, 50, 65, 80, 95, 110, 150, 200, 250, 300, 400, 600, 800, 1600, 2400, 3200 saturated remoulded loess 15, 30, 50, 65, 80, 100, 130, 150, 180, 200, 400, 600, 800, 1600, 3200 saturated remoulded filling 15, 30, 50, 65, 80, 95, 110, 150, 200, 400, 600, 1600, 2400, 3200

Tab.2 Program of oedometer tests

Tab.3

Tab.3

Tab.3 Program of triaxial stress path tests test types saturated natural loess saturated natural filling saturated remoulded loess saturated remoulded filling conventional triaxial compression test CTC50kPa CTC100kPa CTC50kPa CTC100kPa CTC200kPa CTC200kPa CTC200kPa CTC200kPa CTC400kPa CTC600kPa CTC600kPa CTC600kPa CTC600kPa triaxial compression test TC50kPa TC100kPa TC50kPa TC100kPa TC200kPa TC200kPa TC200kPa TC200kPa TC600kPa TC600kPa TC600kPa TC600kPa reduced triaxial compression test RTC50kPa RTC100kPa RTC50kPa RTC100kPa RTC200kPa RTC200kPa RTC200kPa RTC200kPa RTC600kPa RTC600kPa RTC600kPa RTC600kPa

Tab.3 Program of triaxial stress path tests

Table 3 shows the program of triaxial tests. Firstly the saturated sample was saturated again by back pressure until the B-value reached 0.98, and then was consolidated at confining pressure. After the consolidation was completed, undrained stress path test was taken, as shown in Fig.7. The axial displacement rate was 0.033 mm/min in conventional compression test (CTC) and the loading rates of deviatoric stress q˙ in triaxial compression test (TC) and reduced compressiontest (RTC) can be determined by Eq. (1).

where qCTCpeak is the peak deviatoric stress of CTC test at the same confining pressure; 364 is the shear test time of CTC test, the unit is minute.

Fig.7

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.7 Three stress paths in triaxial tests

Figure 8 presents compression curves of the tested soils in oedometer tests and isotropic consolidation lines of the tested soils in triaxial tests. In oedometer test, p' = ( σ₁+2 σ₃)/3, σ₃ = (1-sin ϕ') σ₁, ( ϕ' is shown in Table 8). Figure 9 shows e-log p curves in oedometer tests, where p is vertical pressure. Table 4 shows consolidation yield stress determined by Casagrande method [ 31]. Traditionally remoulded samples consolidated from slurry have no consolidation yield stress [ 32]; while from Fig. 9 and Table 4, remoulded saturated loess and remoulded saturated filling in this test have consolidation yield stress. The reason is that compaction causes the pre-stress during preparation of remoulded soil. Table 5 shows compressibility of the tested soils. The results show that loess is more compressive than filling, which is caused by higher void ratio of loess; the natural soil is more compressive than the remoulded soil when vertical stress exceeds consolidation yield stress, which is caused by breakage of interparticle bonds during compression for natural soil [ 33]. As shown in Table 5, the gradient of K₀ consolidation line is almost the same as that of isotropic consolidation line for the remoulded soil, which agrees with the principle of Cam clay model [ 34], while gradient of K₀ consolidation line is some different from that of isotropic consolidation line for the natural soil, which indicates the anisotropy of consolidation behaviour for the natural soil. As shown in Fig. 10, the distribution of major axis of particles displays anisotropy for natural soils and remoulded soils. But during consolidation, the anisotropy can be destroyed more easily for the remoulded soils than that for the natural soils due to interparticle bonds and fabric in the natural soils.

Tab.4

Tab.4

Tab.4 Consolidation yield stress of samples samples consolidation yield stress/kPa saturated natural loess 79 saturated natural filling 315 saturated remoulded loess 29 saturated remoulded filling 53

Tab.4 Consolidation yield stress of samples

Tab.5

Tab.5

Tab.5 Compressibility of natural soils and remoulded soils soil types saturated natural loess saturated natural filling saturated remoulded loess saturated remoulded filling compression coefficient α1-2/MPa-1 1.11 0.17 0.82 0.55 compression index cc 0.34 0.25 0.28 0.17 gradient of K0 compression line λ 0.152 0.113 0.125 0.073 gradient of isotropic consolidation line λ* 0.199 0.088 0.132 0.072 Note: * calculated by the value at consolidation pressure 200kPa and 600kPa.

Tab.5 Compressibility of natural soils and remoulded soils

Fig.8

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.8 Compression curves in oedometer test and consolidation point in triaxial test

Fig.9

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.9 e-log p curves in oedometer test

Fig.10

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.10 Distribution of major axis of particles of a region observed on vertical plane of sample. (a) Natural loess; (b) remoulded loess; (c) natural filling; (d) remoulded filling

The effective stress paths and the paths of total stress minus static pore pressure [(T- u_s)SP] of the tested soils are shown in Figs. 11-14. u_s is the back pressure, which is used to improve degree of saturation. Strain-softening behaviour is accompanied by the decrease of mean stress for TC stress path, which means the servo of system can not keep mean stress constant. Strain-softening behaviour is accompanied by the decrease of axial stress for RTC stress path, which means the servo of system can not make axial stress constant. Hence the stress path is not exactly TC stress path or RTC stress path after peak deviatoric stress. Because axial stress is controlled by displacement in CTC test, the stress path is also CTC stress path after peak deviatoric stress.

Fig.11

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.11 Test results of saturated natural loess. (a) Effective stress paths and critical state line in p′- q space; (b) (T- us) stress path in p- q space; (c) deviatoric stress versus axial strain curves; (d) excess pore pressure versus axial strain

Fig.12

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.12 Test results of saturated remoulded loess. (a) Effective stress paths and critical state line in p′- q′ space; (b) (T- us) stress path in p- q space; (c) deviatoric stress versus axial strain curves; (d) pore pressure versus axial strain

Fig.13

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.13 Test results of saturated natural filling. (a) Effective stress paths and critical state line in p′- q space; (b) (T- us) stress path in p- q space; (c) deviatoric stress versus axial strain curves; (d) pore pressure versus axial strain

Fig.14

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.14 Test results of saturated remoulded filling. (a) Effective stress paths and critical state line in p′- q space; (b) (T- us) stress path in p- q coordination space; (c) deviatoric stress versus axial strain curves; (d) pore pressure versus axial strain

Tab.6

Tab.6

Tab.6 Failure types of tested soils samples strain-hardening strain-softening saturated natural loess CTC600kPa CTC50kPa, CTC200kPa, TC400ka, TC50kPa, TC200kPa, TC600ka, RTC50kPa, RTC200kPa, RTC600ka saturated natural filling CTC50kPa, CTC200kPa, CTC600ka, TC50kPa, TC200kPa, TC600ka, RTC50kPa, RTC200kPa, RTC600ka saturated remoulded loess CTC600kPa CTC100kPa, CTC200kPa TC100kPa, TC200kPa, TC600ka, RTC100kPa, RTC200kPa, RTC600ka saturated remoulded filling CTC100kPa, CTC200kPa, CTC600ka, TC100kPa, TC200kPa, TC600ka, RTC100kPa, RTC200kPa, RTC600ka

Tab.6 Failure types of tested soils

The types of stress-strain relations include strain-hardening and strain-softening. Table 6 lists the types of stress-strain relations. Soils under RTC stress path are more likely to behave strain-softening than soils under TC stress path, and soils under CTC stress path are most difficult to behave strain-softening. The natural soils are more likely to behave strain-softening than the corresponding remoulded soils.

The comparison between remoulded loess and that of natural saturated loess on shear behaviour is shown in Fig. 15. Void ratio of remoulded loess is lower than that of natural loess at every consolidated pressure. The results show that the contribution of structural bonds to strength is bigger than the contribution of void ratio at low confining pressure, so the strength of natural loess is higher than that of remoulded loess. However, at high confining pressure, structural bonds would be damaged by confining pressure and the contribution of void is bigger than that of structure bonds, hence the strength of remoulded loess is higher than that of natural loess.

Fig.15

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.15 Comparison of shear behavior between saturated remoulded loess and saturated natural loess. (a) Deviatoric stress versus axial strain curves; (b) excess pore pressure versus axial strain curves

Brittleness index is defined by Eq. (2) [ 35].

where qp is the peak strength and qss is the steady state strength.

Tab.7

Tab.7

Tab.7 Softening coefficient of tested soils (%) consolidated stress/kPa saturated natural loess saturated natural filling saturated remoulded loess CTC TC RTC CTC TC RTC CTC TC RTC 50 27.84 20.54 40.72 21.13 8.12 15.04 22.09 200 13.25 18.90 21.10 13.37 20.42 22.60 5.55 11.15 600 20.15 7.15 7.37 11.47 2.38 14.88

Tab.7 Softening coefficient of tested soils (%)

Brittleness indexes of the tested soils under different stress paths are shown in Table 7. It can be seen that the maximum brittleness index is 41%. This is smaller than that of loose sand or loose silt-sand mixture, which is always above 90% or almost 100% [ 3, 4, 5].

Stress-strain patterns of loose clean sands include stable behaviour at low confining pressure, temporary instability at middle confining pressure and static liquefaction at high confining pressure. For natural loess and natural filling in this study, there is no temporary instability pattern and the soils would behave more stability and the brittleness indexes decrease with increase of confining pressure. The difference is caused by the different consolidation character. The isotropic consolidation line of loose sand is different from that of loess or filling, that is, the void changes little with confining pressure for loose sand and void changes great with confining pressure for loess or filling. At low confining pressure, loose sand contracts firstly and then dilates, hence negative pore pressure is generated resulting in strain-hardening behaviour. At high confining pressure, loose sand contracts all long and positive pore pressure is generated resulting in strain-softening behaviour. While for loess or filling, the void ratio at low confining pressure is significantly higher than that at high confining pressure and the soils at low confining pressure are more likely to contract during shearing resulting in strain-softening.

Table 8 shows the effective strength indexes of tested soils. As shown in Figs. 11-14, the critical effective stress points at failure under the three stress paths almost lie on one line for every tested soil, which indicates critical state is independent of the three stress paths.

Tab.8

Tab.8

Tab.8 Effective strength indexes of tested soils samples c'/kPa ϕ'/(°) saturated natural loess 6.4 30.8 saturated remoulded loess 1.4 34.3 saturated natural filling 3.1 34.2 saturated remoulded filling 2.6 33.5

Tab.8 Effective strength indexes of tested soils

Because there is no strain-softening behaviour for remoulded filling, the TC stress path is TC stress path all long and the RTC stress path is RTC stress path completely. Therefore the difference of total strength indexes under the three stress paths can be investigated strictly. The total strength indexes can be calculated by effective strength index and the pore pressure coefficient. The method predicting total strength indexes under TC and RTC stress path from the test results of CTC test based on unique effective strength index has been given by Ref. [ 36], in which the cohesion is assumed to be zero. Total frictional angle under CTC stress path, TC stress path and RTC stress path can be expressed by Eqs. (3)-(5).

where a is pore pressure parameter of Henkel.

Tab.9

Tab.9

Tab.9 Total strength indexes of saturated remoulded filling samples c/kPa ϕ/(°) CTC 8.7 14.9 TC 5.1 16.8 RTC 6.6 26.8

Tab.9 Total strength indexes of saturated remoulded filling

Tab.10

Tab.10

Tab.10 a and Af in test for saturated remoulded filling test types a Af CTC100kPa 0.77 0.26 CTC200kPa 1.32 0.44 CTC600kPa 1.39 0.46 TC100kPa 1.34 0.45 TC200kPa 1.54 0.51 TC600kPa 1.64 0.55 RTC100kPa 1.64 0.55 RTC200kPa 1.54 0.51 RTC600kPa 1.79 0.60

Tab.10 a and Af in test for saturated remoulded filling

Tab.11

Tab.11

Tab.11 Predicted total strength indexes of saturated remoulded loess samples ϕcu/(°) TC 18.9 RTC 34.8

Tab.11 Predicted total strength indexes of saturated remoulded loess

Table 9 shows the measured total strength indexes of remoulded saturated filling. Table 10 presents the values of a in the tests at failure for remoulded saturated filling. According to Eq. (3), the value of φ′ and ϕcuCTC, the a can be calculated and the value is 1.5. Table 11 shows the predicted total strength index under TC stress path and RTC stress path according to Eqs. (4), (5) and a=1.5. It can be seen the predicted value in Table 11 is close to the measured value in Table 9. The results show that the total friction angle under RTC stress path is the largest and the total friction angle under CTC stress path is the smallest.

Test results show that pore pressure under CTC stress path is the largest and the pore pressure under the RTC stress path is the smallest in the three stress paths at the same confining pressure. The pore pressure is negative firstly and then becomes positive for RTC stress path. The maximum deviatoric stress appears later than the minimal pore pressure for RTC stress path. The pore pressure behaviour is different from that of clays under RTC stress path. For clays, the pore pressure is positive firstly and then becomes negative with increase of strain [ 6, 7].

Figures 16 and 17 show the isotropic consolidation line (ICL) and critical state line (CSL) of every tested soil in e-log p′ space. As for remoulded saturated soil, the critical state line is almost a straight line, which means that the stress path does not influence the critical state. As for natural saturated soil, there is a turning point on the CSL, which is similar to the ICL. Hence, the difference between natural soil and remoulded soil in e-log p′ space is that the CSL is almost straight for remoulded soil and the line has a turning point for natural soil.

Fig.16

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.16 ICL and CSL of saturated natural soils in e-log p′ space

Fig.17

	Figure Option ViewDownloadNew WindowDownload As Powerpoint Slide
	Fig.17 ICL and CSL of saturated remoulded soils in e-log p′ space isotropic consolidated lines

The critical state of natural loess or remoulded loess in CTC600kPa test lies far from the CSL defined by other tests. The reason maybe that the soil in CTC600kPa test contains low void ratio after consolidation and behaves like clay, while soils in other tests behave like silt.