An attempt to monitor pore pressure changes in a block sample during and after sampling

Sampling of sensitive clays is a challenging task as their engineering properties, such as undrained shear strength, stiffness and preconsolidation stress, are easily affected by sample disturbance. Sample disturbance in such materials is primarily caused by

• borehole drilling (e.g. Hvorslev, 1949; Clayton, 1986)

• sampler type (e.g. Berre et al., 1969; Lefebvre & Poulin, 1979; La Rochelle et al., 1981; Baligh, 1985, Amundsen et al., 2015a)

• sealing, transportation, thermal variations, storage method, trimming and handling during preparation for testing (e.g. Bozozuk, 1971; Arman & McManis, 1976; La Rochelle et al., 1976, 1986; Atkinson et al., 1992)

• stress relief during and after sampling (e.g. Ladd & Lambe, 1963; Skempton & Sowa, 1963; Noorany & Seed, 1965; Bjerrum, 1973)

• changes in the physicochemical properties (e.g. Torrance, 1976; Lessard & Mitchell, 1985).

Over the years, significant development has taken place to overcome sample disturbances. Extensive studies have shown that block sampling is among the best methods of collecting high-quality samples of soft clays (e.g. DeGroot et al., 2005; Karlsrud & Hernandez-Martinez, 2013). Accordingly, it has become more common to use Sherbrooke block sampling (Lefebvre & Poulin, 1979) in sensitive clays as this method ensures that the soil remains unaffected by shear distortions during sampling (e.g. Lacasse et al., 1985; Tanaka et al., 2001). However, one of the issues that is challenging with block sampling is the stress relief. Despite careful handling and transportation, the clay samples may exhibit poorer quality than anticipated – especially for low-plasticity sensitive clays. The effect of stress relief is exemplified in Fig. 1 using several samples from seven sites in Central Norway. The results indicate a deterioration in the quality of samples with increasing stress relief. Since the development of stress relief over time is of high importance, an attempt has been made to study this by continuous monitoring of the pore pressure inside a sample during the sampling process. Before sampling, a wireless pore pressure sensor was inserted into the ground and left to stabilise. A block sample was taken of the soil containing the piezometer and the pore pressure was recorded throughout the insertion, sampling, handling and storage.

During tube sampling, the soil experiences load changes that follow a compression–extension–compression strain cycle (Baligh et al., 1987). This is eliminated during block sampling. Unfortunately, a block sample experiences the effect of stress relief, which may lead to swelling of the soil structure. After a block sample is taken out of the ground, it develops a negative pore water pressure (Ladd & Lambe, 1963). This generates a pressure gradient, which transports the pore water from the remoulded and destructured clay on the surface of the sample to the intact clay in the middle (Kallstenius, 1971). This water migration may be accompanied by gas exsolution if the pore water contains dissolved gases (Fredlund et al., 2012). As a result of this, the pore volume is allowed to expand, which leads to a reduction of the residual effective stress (RES or p′_r) and causes the intact soil to swell (e.g. Tanaka & Tanaka, 2006).

Table 1 presents an overview of the stress changes a saturated block sample experiences, from an initial stress condition before the sampling (a), during sampling (a–b), and after (b–c). During sampling, a block sample is removed from its in-situ conditions into an isotropic state where it develops a negative pore water pressure. Following Ladd & Lambe (1963), the pore pressure after the sampling should be

where A_u = (Δu − Δσ_h)/(Δσ_v − Δσ_h). Ladd & Lambe (1963) found A_u for unloading of soil samples to be between − 0·1 and 0·3. However, based on the triaxial tests conducted on the Tiller clay, an elastic response with A_u = 1/3 is assumed

Note that K′₀ has a large influence on the theoretical value of u_ps. For an isotropic stress state, u_ps = − σ′_v0, whereas for an anisotropic stress state (K′₀ = 0·5) it is equal to − 2σ′_v0/3.

The current design of the block sampler does not restrict a sample from swelling during and after sampling, and therefore, the loss of RES will increase with time. However, materials of low overconsolidation ratio (OCR) and I_p may not follow a theoretical path during unloading, (a–b) as in Table 1, and rather follow the (a–d) path where p′_r < p′_ps. Similar observations are also reported in the literature (Table 2), where the relation between I_p and the OCR is illustrated (e.g. Gens, 1982; Hight & Burland, 1990; Carrubba, 2000).

The Norwegian University of Science and Technology (NTNU) has established a geotechnical research site at Tiller, near Trondheim. The characterisation and engineering properties of the clay are well documented in the literature (e.g. Sandven et al., 2004; Gylland et al., 2013; Amundsen et al., 2016b) and the geotechnical profile is shown in Fig. 2.

The pore pressure measuring equipment was modified from a commercially produced wireless Micro-Diver datalogger (vanEssen Instruments), shown in Fig. 3. The Micro-Diver consists of a pressure sensor, memory for storing measurements and a battery. The frequency of the measurements is adjustable; in this study, it was set to one measurement every 15 s. The measurements may be downloaded to a computer after the testing is completed. A porous stone filter has been added to protect the membrane and pressure sensor from the soil particles. Before testing, the piezometer calibration was checked using a vacuum pump and a triaxial cell. The water reservoir and the filter were saturated with glycerine prior to installation.

The samples in this study were taken by a downsized version of the Sherbrooke block sampler (160 mm in diameter and 300 mm in height), also called the mini-block sampler, made at NTNU (Emdal et al., 2016).

The installation procedure of the piezometer in the clay is shown in Figs 4(a)–4(c). A hollow steel rod with the piezometer inside was pushed into the ground and the piezometer was released in the undisturbed clay at 10 m. The piezometer was left in the ground (Fig. 4(c)) to stabilise the pore pressure from the installation process. The piezometer was not attached to a cable. Thereafter, the piezometer was retrieved from the ground by overcoring with a mini-block sampler and then sealed immediately (Figs 4(d)–4(f)). The block sample BL0, in Fig. 4(f), contained the piezometer, and only routine tests have been done to compare the material to the reference block samples.

The reference block samples were extracted from a parallel borehole, about 3 m away, BL1 and BL2 in Fig. 4(f). The samples were sealed and immediately transported to the same laboratory where they were tested. BL1 (9·95–10·25 m) was tested in the laboratory 2 h after the sampling and BL2 (10·25–10·55 m) was stored for 48 h prior to testing in the same laboratory.

For comparison, constant rate of strain (CRS) oedometer tests and anisotropically consolidated undrained triaxial compression (CAUC) tests were conducted on two reference block samples. The testing equipment, procedures and operator were the same for all tests. The block samples were sliced in an equal manner, as shown by Amundsen et al. (2016b), and the location of the oedometer and triaxial test specimens within the block was the same, close to the undisturbed centre of the block. The specimens were tested immediately after trimming. In general, human errors were minimised and the tested specimens were practically identical.

Figure 5 shows the measured pore pressure during insertion of the piezometer and sample extraction, which is illustrated in Fig. 4. The drilling started 33 h after the installation, marked as (3) in Fig. 5, where the upper 5 m of the soil has been removed and water added to the borehole. This is an established procedure to prepare the borehole prior to block sampling in Norway (Karlsrud et al., 2013). Point (4) in Fig. 5 represents remoulding (by the auger) of the sensitive clay at a depth of 7–9 m, during which the pore pressure varied between 48 and 130 kPa. Afterwards, the borehole was filled with water to the ground surface, denoted as (5) in Fig. 5. The carving of two dummy block samples to approach the last section above the piezometer started from point (6) in Fig. 5. This has been registered as an increase in pore pressure, which reduces back to u₀. The cutting of block sample BL0, the sample with the piezometer, started from point (7) in Fig. 5.

The variations in the measured pore pressure in Fig. 5 indicate that the total vertical stress in the BL0 sample was not constant during drilling. Due to the complexity of the stress condition, this has not been fully addressed in this paper.

In Fig. 6, a more detailed description of the sampling process and the time after is shown, which corresponds to (7) in Fig. 5. The sample was cut during an 8 min period, (7)–(8) in Fig. 6(a), and lifting followed immediately (9)–(10). During this process, several steel rods were detached from the drilling equipment. This was captured by the piezometer (Fig. 6(a)) as short time delays before the lifting continued. A small pore pressure dissipation of 0·4–0·8 kPa occurred during these time delays. In total, the cutting and lifting of the block sample took about 17 min. Figure 7(a) shows the mini-block sample right after the sampling, with cutting debris covering parts of it. Figure 7(b) shows the sample after the sealing.

Figure 6(b) illustrates the dissipation of the residual pore pressure after sampling (10) and sealing (11), and during the 38 h storage before it was transported 12 km (12)–(13) to the laboratory where it was opened (Fig. 7(c)).

An initial stress condition is presented in Fig. 8(a). Figure 8(b) shows the pore pressure measurements inside the block sample during sampling and lifting of the block in the borehole. During the carving of the block, the total stress changes from p₀ to p_mud. After sampling and during lifting, the sample is submerged in drilling mud, which gives a total isotropic stress (p_mud) on the sample. The difference between the stress from the mud and the measured pore pressure inside the block is the RES in the centre of the sample.

The lowest value of residual pore pressure observed in the block sample was − 13·7 kPa during the first 10 min after sampling. From this point onwards, the sample started to lose its residual pore pressure and within 6 h after sampling it was –4 kPa.

Results from two oedometer and two triaxial tests from the reference block samples are presented in Fig. 9, along with an interpretation in Table 3. The oedometer tests in Fig. 9(a) show that the sample which was tested after 48 h of storage (CRS-2) had about 4% lower preconsolidation stress and 20% lower stiffness than the sample tested 2 h after sampling (CRS-1). The normalised undrained shear strength in the triaxial test CAUC-2 decreased by 5%, the strain at failure remained unchanged and the pore pressure increased slightly. These results indicate that the material parameter values tend to decrease slightly during the storage period. According to the literature (Ladd & Lambe, 1963; Hight & Leroueil, 2003), this cannot be explained by the decrease of the RES. Other reasons such as inhomogeneity of the material, variations during the sampling process, the sealing technique and transport could have caused or attributed to the discrepancies in the measured values. The more realistic in-situ response that has been observed in the samples with little storage time underlines the benefit of beginning testing as soon as possible after sampling. In this case, even sooner than 2 h after sampling would have been preferred. However, due to the limitations of this study more investigations are needed before conclusions can be drawn.

The theoretical value of the RES after the unloading of the sensitive Tiller clay is p′_ps = − u_ps = 65 kPa, with an assumed K′₀ = 0·5 (based on Brooker & Ireland, 1965; Gylland et al., 2013). The observed maximum RES (p′_r = 13·7 kPa when the sample is above ground) is very low compared with a theoretical value of 65 kPa and p′_r/p′_ps = 0·21 and p′_r/σ′_v0 = 0·14. The fact that the negative pore pressure generation was quickly dissipated confirms that the sample does not develop a high RES due to swelling and water migration during the unloading in the borehole. A rough estimation of the time (t) it would take for the clay to dissipate the RES has been calculated as follows

where T_v is the time factor (assumed T_v = 1); c_h is the coefficient of horizontal consolidation; c_v is the coefficient of vertical consolidation (assumed c_h = 5c_v) and H is the drainage length, which is equal to the radius of the block sample (R = 8 cm) (Terzaghi et al., 1996). An average c_v for the sensitive Tiller clay is 30 m²/year. The dissipation time is about 20 min. The piezometer measurements agree with the estimation. Other sources such as inhomogeneity, low plasticity and high permeability of the Tiller clay could have attributed to a low RES to some degree. However, no pure silt layers have been observed near the piezometer and the soil was a clay with varying plasticities, similar to Fig. 10.

Schjetne (1971) conducted an in-situ test on sensitive clay of low plasticity, I_p = 3%. He measured the pore pressure during and after sampling with a hypodermic needle in a 95 mm tube sampler. The results are compared with the measurements in a block sample in Fig. 11. The results show that the tube sample goes through a compression–extension–compression strain cycle, (1)–(3) in Fig. 11, which may have resulted in an excess pore pressure five times that of the initial. However, the pore pressure variation in the block sample was insignificant during the same process. Also, during the drilling of the borehole (Fig. 5), the pore pressure variations were minor. This confirms that block sampling is a much gentler technique compared with tube sampling. The final ratio of the p′_r/p′_ps values were 0·27 for the tube sampler and 0·21 for the block sampler, (4) in Fig. 11.

Table 2 is a compilation of results for other materials from the literature, which is compared with the Tiller clay RES. Varieties of natural materials, collected using tube and block samplers, as well as reconstituted samples are included. It is clear that p′_r/p′_ps and p′_r/σ′_v0 are much lower than the theoretical values of 1·0 and 0·67–1·0, assuming an elastic response. A high-plasticity Onsøy clay was able to maintain its RES (6–15 kPa) for at least 3 months, where the p′_r/σ′_v0 ratio varied between 0·07 and 0·15. Other examples are the two medium plasticity clays, Lierstranda and Ishinomaki, which exhibit a p′_r/σ′_v0 ratio of 0·01–0·11. There are, however, very few data on natural low-plasticity clays, with an in-situ measurement performed by Schjetne (1971) being, to the best of the authors’ knowledge, the only test conducted on such a material. These observations, including other data from natural clays and the conducted in-situ measurements in this study agree. The reconstituted clays are seen to generate a higher p′_r/σ′_v0 ratio compared with natural clays.

There are several limitations to this study. First of all, there was only a single in-situ test conducted, and only two reference samples were sampled and stored. Furthermore, only a limited number of CRS and CAUC tests has been conducted. To strengthen the findings reported herein, more tests are required. Another challenge is the inhomogeneity of the sensitive clay deposit, meaning that the material parameters vary slightly in the sample, as shown in Fig. 10. Finally, the volumetric change of the block sample during sampling, storage and test set-up was not measured and is therefore unknown. Work is currently ongoing in reducing these limitations.

This study was conducted to monitor the pore pressure variations during and after sampling in a low-plasticity sensitive clay. It was observed that the sample developed less RES compared with theoretical values. Laboratory testing on the reference samples indicated that the quality of the sample tends to deteriorate with increasing storage time.

The pore pressure response was compared with a field test conducted by Schjetne (1971) on low-plasticity clay. Despite the complete loss of RES, the block sample results indicate very good sample quality. However, sample quality deterioration in the first hours after sampling has been demonstrated, emphasising the importance of short storage time.

From this study, the following remarks may be made.

• Reduction of the RES begins to take place less than 10 min after sampling.

• Sealing of the block sample is important to prevent the sample from swelling and to maintain its RES.

• The reduction of the RES and swelling may yield poorer quality samples. Therefore, testing of the sample should be conducted as soon as possible.

In this work, a preliminary attempt was made to study pore pressure changes during and after sampling. To give a robust recommendation of RES after sampling, more in-situ testing should be done on various materials and stress conditions.

Engineers G. Winther, E. Husby, P. Østensen, F. Stæhli, T. Westrum, K. I. Kvisvik and E. Andersen at the NTNU Geotechnical Division are acknowledged for their skills and knowledge that made the experimental work possible. The authors also extend their thanks to Multiconsult AS and NGI for their help. Master students at NTNU, H. Dang and H. Tovslid are acknowledged for some of the laboratory testing. The authors thank Dr S. Degago from Norwegian Public Road Administration for his comments on the paper. The authors also acknowledge support from the inter-governmental research program Natural hazards: Infrastructure, Floods and Slides (NIFS, 2012–2015). This work was supported by the Research Council of Norway, grant no. 246629.