Triaxial creep tests of glacitectonically disturbed stiff clay – structural, strength, and slope stability aspects

4.1 Soil structure recognition

The preliminary assessment of the soil structure before its removal from Shelby samplers was performed using X-ray densitometry. Using this method, the soil inside two metal samplers was tested. The passage of X-rays through the thin metal walls of the sampler caused a constant error in registering the soil material density value; however, a qualitative characteristic of soil density variation was obtained without the risk of structural changes during the extrusion of the soil material from the sampler. This allowed indicating the zones of structure disturbance along the core. The obtained densitometric images had low resolution (pixel size ∼1 mm), so the boundaries of zones with various densities were blurred. On the other hand, in a short time of exposure (∼30 min), it was possible to identify hidden zones of the loosened material and select the appropriate parts of the core for cutting cylindrical samples (h = 72 mm in height and d = 36 mm in diameter) oriented with a vertical axis in accordance with the natural orientation of the soil substrate.

Then, the samples for strain and strength tests in the TRX apparatus were subjected to microtomographic imaging according to the methodology in refs [23,24,25]. A MicroXCT-400 model, produced by the American company Xradia, was used. X-ray high-resolution computed microtomography tests (XµCT) included the following: data recording at different directions of exposure, their assembly as several series of projections, image processing, and final three-dimensional visualization. The exposure time of a single image was 6 s. A series of 1,024 X-ray images of each sample (in the format of 8-bit BMP images with a resolution of 1,024 × 1,024) was obtained.

The obtained images showed the variability of the coefficient of X-ray absorption, influenced mainly by the density variation of the solid phase of the clay material. The most compacted areas were characterized by lighter shades, while the darkest areas were recognized as gaps, often filled with loose materials.

During the exposure, the soil samples were protected with a rubber membrane, which eliminated the possibility of water evaporation. Due to the low density and membrane thickness in comparison to the size of the soil sample, it did not significantly affect the quality of radiographs. With the assumed cylindrical dimensions of the samples, pixels with a side of ∼40 µm were obtained. To assess the possibility of imaging quality improvement, several additional exposures of smaller, centimeter fragments of clay samples were carried out. The imaging of structural discontinuities affecting the deformation of the soil was therefore fully sufficient when using the above-mentioned exposure parameters. On the other hand, the ranges of pore size of the analyzed clays, determined by SEM, often presented in the literature, were several dozen times lower (0.75–1.90 µm; ref. [2]). The SEM method, due to the small size of the scanned surfaces and obtaining 2D images, is not as comprehensive in the assessment of structural discontinuities as the presented microtomographic imaging.

4.2 Shear strength tests

Shear tests have been performed with the use of triaxial systems, based on computer control and data acquisition by GDS Instruments equipment and software solutions. A general methodology of triaxial tests was consistent with current standards [26], and state-of-the-art practices were described in refs [27,28], with some unique addition.

Preparation of test samples for both standard shear strength determination (the names of the samples S₁, S₂, S₃ referred to the first letter of “shearing”) and tests with additional consideration of the creep stage was preceded by uniformly adopted saturation conditions.

According to the commonly known and established practice, preparation of the samples included saturation by backpressure technique [29] and isotropic consolidation. Backpressure value used to achieve demanded saturation level (the end saturation criteria were B values, which were ∼0.84, which refers to the research of ref. [30] was up to 800 kPa. Effective consolidation stresses applied to the samples in one series were σ ′ 3 = 100 , 200, and 300 kPa (the consolidation took up to 3 days). These values represent the general stress state, water level depth, and depth of sampling.

The samples were subjected to appropriate strength tests at increasing stress σ ₁ automatically adjusted to the assumed, previously calculated as appropriate, constant displacement rate vs = 0.01 mm/min.

These studies were conducted under undrained conditions (CU). This test type has been chosen, as representing a more critical scenario of subsoil conditions, giving insight into pore pressure behavior, resulting in effective strength parameters, and also being more time efficient in comparison to the CD test. The deformation increase observed in the CU tests depends on the load as well as on soil reaction by changes in soil porosity, without the parallel participation of filtration and consolidation processes. On the other hand, in the case of CD tests (drained), these processes cooperate, causing changes in the character of deformability and strength. The significance of these factors has been repeatedly assessed by analyzing various results of drained and undrained triaxial tests. For example, in normally consolidated clays from the Macao region of China, ref. [31] obtained higher deformation values and higher soil strengths from drained tests than in undrained ones. These effects were associated with the influence of filtration and consolidation processes. Therefore, the analysis confirmed that the CU were more unfavorable and, therefore, much more safe in engineering predictions.

The effective values of internal friction angle and cohesion were determined according to classical methodology resulting from the Coulomb–Mohr theory [32]. For the representation and analysis of the results, standard MIT stress path parameters [33] have been used, i.e., mean effective stress s ′ = ( σ ′ 1 + σ ′ 3 ) / 2 and shear stress t ′ = t = ( σ ′ 1 − σ ′ 3 ) / 2 , which is half of the deviatoric stress q. The maximum value of the deviatoric stress was considered as shear stress criterion.

The obtained values of the internal friction angle were also compared with the shear angles (θ) of the samples determined after the shear tests completion based on the analysis of the directions of the shear surfaces in the microtomographic images. The theoretically specified shear angle in a homogeneous material should be θ = 45° + (ϕ′/2). Deviations from this principle may indicate the heterogeneity of the soil material, as well as structural features as microcracks and sedimentation lamination [34,35].

Other results of the study of overconsolidated clays from the Lower Silesia region [35] also showed lower than theoretically expected values of θ. These effects were explained by reactivating discontinuities or weak zones in the soil structure. Their presence in some samples resulted in decreasing the shear angles in comparison to the theoretical ones. This comparative approach has also been used in interpreting the research presented in this article.

4.3 Triaxial creep tests

The creep tests conducted on the second group of samples (the names of the samples C₁, C₂, C₃ referred to the first letter of “creep”) were planned with the stress level (SL) = q/q _f = 0.60, i.e., 60% of the expected maximum deviatoric stress (q _f). Then, changes in the axial deformation (ε ₁) were observed for 14 days as a function of time, and then, the load was continuously increased until the failure was achieved. The choice of the vertical stress value at which creep observations were made may be optional, but usually with the recommendation that creep should be declining, i.e., it does not lead directly to loss of soil strength. The value of SL = 0.6 adopted in this study was determined on the basis of the results of authors’ experiments [23] and the works of ref. [36], where decelerating creep was observed at stresses 30 to 70% of the maximum deviatoric stress ( σ ′ 1 − σ ′ 3 ) . Triaxial stress is an important aspect of creep studies [37]. In the following research, the expected values of the maximum deviatoric stress at failure were determined based on the results of standard triaxial CU tests of the first group soils that had the same lithogenetic type and structure. The maximum deviatoric stress in the tests of the second group soils (q _max) obtained after the completion of the creep stage and re-application of the constant load increase may differ from the reference results from the soils of the first group (q _f). It was conditioned by the natural structural differentiation of the studied soils as well as by the influence of the previous creeping. The analysis of these differences is one of the goals of research and analysis.

The creep of the investigated soils was characterized based on changes in the increment of axial deformations ε ₁ as a function of time t. Resulting from the analysis of the equation (ε ₁) − log t, three qualitative types of deformation changes might be distinguished: decelerating, quasi-steady state, and accelerating. Hence, for the quantitative characterization of the creep behavior in any time interval, a dimensionless parameter m was adopted [38].

The parameter m is based on the model assumption of logarithmic linearization of the increment in axial deformation as a function of time. The decreasing velocity of deformation corresponds to the value of m > 0 (decelerating creep). As the load increases, the soil shows a constant rate of deformation, which corresponds to the value of m = 0. The value of m < 0 corresponds to the increase in the rate of deformation that occurs directly before shearing.

Figure 7 shows a scheme for determining m value, comprising, in turn:

1. Registration of vertical deformations as a power function of time e = at ^b , if a > 0 and 0 < b < 1, the deformation has a decelerating character (Figure 7.1).

2. A power approximation of the deformation velocity versus time, at a decelerating creep a > 0 and b < 0 (Figure 7.2).

3. Logarithmic linearization of deformation velocity versus time (Figure 7.3).

4. The relationship between the types of deformation changes: decelerating, quasi-steady state, and accelerating, and the m values: positive, zero, and negative (Figure 7.4).

Figure 7

Calculation scheme of parameter m and its various values in relation to the character of the strain rate increase.

When discussing the determining m parameter, it is necessary to explain the assumption of the axial strain rate ε ₁ as the creep indicator, despite the triaxial conditions of changes in stress and strain. Ref. [39] showed that the m parameter values calculated on the basis of ε ₁ were close to those determined based on the volumetric deformations (ε _V = ε ₁ + 2ε ₃). At the same time, ref. [40] noted greater uncertainty related to the possible use of ε _V than ε ₁.

5.1 Structural evaluation of tested samples

The purposes of densitometry and computed microtomography research were as follows:

• The assessment of the character and the origin of structural discontinuities of tested soil;

• The selection of samples with similar structural characteristics for comparative rheological and strength tests;

• The assessment of the impact of loads applied in triaxial tests on forming the new structural effects.

The results of low-resolution exposures performed in the densitometer allowed to obtain the preliminary assessment of the structural condition of the cores inside Shelby samplers. The observed disturbances may result from both the drilling and sampling techniques and the natural genetic features of the soil. The dark zones visible in the density images indicated a significant relaxation and even voids (Figure 8). These parts of the core (intervals: 9.12–9.20 m and 10.65–10.95 m below the ground level) were eliminated from further research. Taking the remaining material into consideration, cylindrical soil samples with standardized dimensions (36 mm × 72 mm) were prepared. Before placing them in the triaxial apparatus, microtomographic exposures were performed. This way, spatial data with ∼25 times (and in single XμCT scans even ∼60 times) better resolution were obtained. The possibility of creating any digital cross sections on their basis enabled a multi-directional analysis of the structural variability of the material selected for further research.

Figure 8

Interpretation of the densitometer imaging results of the analyzed clay in Shelby probes.

Structural images of soil samples for TRX tests indicated the diverse nature of the recorded heterogeneities. For example, the image of the U₁ sample (the name refers to the sampling sites, i.e. U as upland) showed extensive, darker fragments of lower density, corresponding probably to sandy inclusions (Figure 9). This indicated the variability of the genesis and transfer of the material even at a small scale. On the other hand, microcracks and voids found in the sample U₂ (Figure 9) were considered to be related to the complex stress state due to glacitectonic strains. Some cracks may show up and develop additionally during soil sampling and/or during extruding it from the sampler and preparing cylindrical samples. A similar genesis of discontinuities was observed in sample V₁ (the name refers to the sampling sites, i.e. V as valley; Figure 9), collected from Powiśle [23], where in V₁ sample, the cracks with an opening of up to several millimeters and a few centimeters long were observed. These examples showed that structural disturbances may appear at various stages of sediment formation and transformation as well as during the collection and preparation of samples for testing. The character and conditions of structural variability may significantly affect the separate, individualized deformation and strength behavior of samples. Hence, the presented samples U₁, U₂ were not qualified for further triaxial tests, as they did not meet the condition of structural similarity to other samples designated to shearing tests series (S₁, S₂, S₃) and creep test series (C₁, C₂, C₃).

Figure 9

The compilation of various images of the Neogene clay from the area of Warsaw slope.

Exemplary microtomographic sections of the chosen for further triaxial tests samples (V₂, S_2, and C₁) are shown in Figure 9. They illustrate typical images of the internal structure, i.e., a relatively homogeneous clay material with a dispersed content of silty fraction, without microcracks visible at the resolution used. Among other structural features, it is worth mentioning that the “nest” clusters with lighter shades indicate a higher density of the soil material. They may indicate the presence of iron compounds (with higher density in relation to the clay matrix). Such effects were found in small sandy enclaves accumulating the mineralized liquid phase in the pore space. The presented interpretations were verified and confirmed directly by cutting the samples after performing triaxial tests and repeated microtomographic images. Characteristic aggregates of sand grains with a rusty color suggesting the presence of iron ions in the third oxidation state were identified.

The examples of the images of the microstructures of Neogene clays after the TRX CU tests (Figure 10) clearly reveal both the shear surfaces and the zones of plastic displacement inside the samples. This indicates a mixed, brittle-plastic mechanism of deformations.

Figure 10

Exemplary images of the post-failure structure of Neogene clay samples (after triaxial tests).

A similar character as well as lithological and structural changes was noted in microtomographic images of samples subjected to the creep stage and then shear (C₂ – Figure 10).

The complementary use of densitometry and microtomography indicates their universal, innovative application in initial quality recognition of full-size soil core and subsequent, detailed structural analysis.

5.2 Shear strength

The shear strength (Table 1) shows some differences when comparing the two groups of test samples with the standard continuous load increase (samples S₁, S₂, S₃) and with a gap during continuous loading, at which long-term creep at constant stress was monitored (samples C₁, C₂, C₃).

Table 1

Characteristics of the course and results of shear strength tests of Neogene clay samples

The values in grey refer to the results of standard TRX CU tests preceded with analogous consolidation conditions.

The underlined angle values indicate the dominant failure plane.

^The use of weights.

The summarizing data and the obtained results are presented in Table 1 characterizing:

• Conditions of initial consolidation;

• Axial deformations and stresses s′ and t (deviatoric) at failure;

• Cohesion and internal friction angle;

• Shear angles determined on the basis of the tomographic images.

In general, despite careful selection of structural similarity, the tested samples showed some differentiation in behavior. It was confirmed by the random variability of the consolidation coefficient in the range (0.6–5) 10⁻⁹ m²/s. It was probably conditioned by local changes in water content and pore pressure distribution.

Figure 11 shows the results of all shear stages of the tested samples using graphs of principal effective stress ratio ( σ ′ 1 / σ ′ 3 ) against the axial strain and pore pressure changes. After the 14-day creep stage, the maximum deviator was most often achieved faster than in the standard triaxial tests. In comparison to the standard tests, the samples of the “C” series show an increase of s′ and t values at failure. For the sample consolidated at σ ′ 3 = 100 kPa, the increase of s′ was 32% and the deviatoric stress t was 72%. For σ ′ 3 = 200 kPa, an increase was as follows: s′ = 23% and t = 36%, while at σ ′ 3 = 300 kPa, 24 and 20%, respectively. Linear strength envelope interpretation of “C” test gives as a result higher angle of internal friction (20% increase comparing to “S” test) and cohesion reduced to zero (Table 1). A possible explanation of the reduction of cohesion is creep-reactivated microdiscontinuities related to the glacitectonic history of the sediment. Due to a limited number of samples available, authors cannot draw firm conclusions, although a very thorough examination of the tested material prior to the test gives confidence toward strength parameters obtained.

Figure 11

Comparison of the shear stage of two exemplary triaxial tests: standard shearing (S1–S3) and taking into account earlier creep (C1–C3).

The variability of the cohesion and the internal friction angle of Neogene clays from Warsaw maybe quite significant [2] – Figure 12, which results from the structural conditions related to the specific sedimentation and glaciotectonic history. This is typical for Polish Neogene clay sedimented in various places like the Bydgoszcz region where its variability is even higher [41]. The soil structure with fissures determining soil strength can be observed also in similar London clay [19,34]. The samples analyzed in this article were characterized by high values of internal friction angle, similar to the upper estimates for the Neogene clays from Warsaw and relatively low values of cohesion, even reduced to zero in the soil weakened by the creep stage.

Figure 12

Failure envelopes for two series of triaxial compression tests (standard and including creep stage) against the range of archival test results.

Comparison of the character of deformations and the failure conditions of tested soils with data from other regions allowed to assess the determinants of deformations. As a reference, it can be assumed from ref. [27] that for plastic clays, strain at failure is from 15 to 30%. Hence, in the pre-consolidated formations, these values maybe even five times lower. For example, for the Neogene clays located to the south of Wrocław, a range of strains 3–6% have been reported [42]. From the analysis of glacitectonically disturbed Neogene clays from Warsaw, comparable 1–5% strains at failure were observed. Similar results can be found in other publications [43]. Low values of axial strains at failure were associated with their stiff, pre-consolidated structure, and the probable initiation of shearing along glacitectonically disturbed weak zones and microdiscontinuities.

An important piece of evidence indicating the role of weak surfaces connected with the history of loads on the strength characteristics was the comparisons of shear angles in the samples θ _cr (after tests with creep stage) and θ _sh (after standard TRX CU) with the expected values according to the theoretical Coulomb–Mohr model. The most common values obtained during the research were θ _cr < θ _sh (with the lowest angles from 35 to 9.5°). It proves the structural heterogeneity of the tested material and the role of local microshears in shaping the steady-state states and the conditions for achieving the maximum shear resistance. This explanation corresponds to the abovementioned relations between the microstructural features and the mechanical properties of the soil.

Complementary strength tests extended for creep stage with documentation of microstructural changes is a novel and comprehensive approach. Creep impact on mechanical properties of soil is constantly an explicated issue, where latest studies of stiff clay concern time and stress dependents, e.g., ref. [44].

5.3 The course of clay creep

The evaluation of the creep influence on the strength properties of soils requires a comparison of the course of tests under continuous load increase (“S” tests – shearing on group I samples) with tests where a 14-day creep stage under a constant load was carried out (“C” tests – creep on samples group II). The SL was supposed to correspond to 60% of the maximum (at failure) stress deviator q _f. The values of q _f were assumed on the basis of the shear strength obtained from the “S” tests on the group I samples. The assumed loads were equal only in the range from 45 to 49% of the real value of the load at failure which was achieved later (for group II samples). Thus, the values of q _f obtained from the “C” tests on group II samples turned out to be about 20–30% higher than the results obtained in the “S” tests carried out on group I samples. It was difficult to decide whether the increase in the value of the stress at failure was caused by the structural reconstruction during the decelerating creep. It should be noted that such relations occurred consistently in tests of samples carried out at stresses σ ₃ equal to 100, 200, and 300 kPa carried out on the samples of overconsolidated soils.

Characteristics of the creep course and the rate of deceleration depend on the stress which the soil is subjected to and on the individual structure of the samples. The graphs presented in Figure 13a show that the values of the vertical strains were inversely proportional to the horizontal stresses σ ₃. These effects are connected with the well-known soil mechanics relations between the compressibility in the triaxial and uniaxial deformation state. The decrease in the vertical displacements values was directly related to the limitation of deformations in the horizontal direction caused by the magnitude of horizontal stress applied. This dependence indicates that together with depth, the creep tendency as well as the range of creep deformations preceding soil shearing will be smaller and this refers also to the slopes. This was confirmed by a number of field observations and inclinometric measurements. The one exception may be the presence of visible or hidden structural discontinuities constituting privileged zones of increased deformation of a separate nature.

Figure 13

Results of single-stage creep presented as (a) increment of axial strain as a function of time and (b) strain rate as a function of time in logarithmic scales.

The course of the creep expressed by the graph shape can also be associated with the structural variability. In comparison to other tests, the C₂ sample tested at σ ₃ = 200 kPa showed the highest rate of deformation, which was maintained over several days (Figure 13b). The initial slight deformations did not show a quasi-stabilization, but they systematically developed (up to the highest value observed in the studies considering creep – Table 1), which may result from the natural structural variability, including local changes of state or particle size distribution.

The differences in the creep behavior were reflected in the values of m parameter (Figure 13b). The lowest value of m = 0.64 corresponded to the creep of the above-mentioned C₂ sample. Potentially, in a soil characterized by decreasing values of m parameter, slow but constant displacements on slopes might be expected (i.e., the lowest rate of loss of displacement velocity of the soil of C₂ character among all other samples). If the soil material has no weakening zones, and the load values are about 40–60% of the shearing stress, after the initial significant deformation, the reduction is observed. The characteristics of the creep velocity changes of tested clays confirmed and extended the results of the analyzes provided in ref. [40], where the values of the parameter m < 1 had been obtained for the overconsolidated soils. Nevertheless, the authors [40] underlay the necessity of creep studies on stiff, overconsolidated soils, because of a small number of tests.