You are on page 1of 22
Recent Developments in Soi and Pavement Mechanics, Amekda (ed) © 1997 Bakoma, Rotterdam, ISBN 90 6410.885 1 Sample isturbance effects in soft low plastic Norwegian clay ‘T.Lunne, T.Berre & S.Strandvik Norwegian Geotechnical Institute, Oslo, Norway ABSTRACT: The Sherbrooke block sampler has been used at several clay sites in Norway. Samples down to 22 m below ground level have successfully been obtained. Standard piston sampling with the $4 mm NGI sampler have been obtained at the same sites. 75 mm and 95 mm piston samples have been obtained in some cases. Piezocone test results are also available. Subsequent laboratory testing (CAUC triaxial and CRS ‘cedometer tests) clearly show the superior quality of the block samples compared to the piston samples; this is at least the conclusion to a depth of 16 m where comparative results are available The effects of sample disturbance are especially dramatic for low plasticity clays. The results are described in detail and the soil parameters that are most effected by sample disturbance are identified. The results are used to develop a new criteria for evaluation of sample disturbance. It is also stressed that results of CAUC and CRS tests on piston samples should be corrected to reduce the effects of sample disturbance. Soil parameters as measured on the block samples are used as reference parameters for a new generation of correlations for interpreting CPTU results. 1 INTRODUCTION Oslo. Figure 2 shows a soil profile from Lierstranda. Below a stone and gravel fill about 3 m thick, the Since NGI was founded in 1953, the items of ‘plastic’ clay extends to about 12 m depth where the sampling, laboratory testing and assessment of ‘lean’ Drammen clay is encountered. The water sample disturbance have been given much attention. content, which is close to the liquid limit, decreases Due to important geotechnical problems in the city of from about 43% at 4-5 m depth to 32% at 25 m Drammen, about 35 km south of Oslo, the Drammen depth. The salt content of the pore water decreases clay has been the subject of much of the research. from about 30 g/l at 4-5 m depth to about 20 g/l at Over the years it is the so-called upper plastic 20 m depth. The plasticity index, in the upper clay is Drammen clay (plasticity index, I, = 27%) that has about 20-25% which is somewhat less than the been most investigated; for instance large pioneering ‘plastic’ clay from the centre of Drammen as noted laboratory test programmes have been performed to above. For the lower clay, Ip ranges from 11 to 16% lear more about its cyclic behaviour (e.g, Andersen The undrained shear strength as measured by triaxial €t al, 1980) since this clay has about the same tests on specimens from tube samples consolidated Plasticity index as many North Sea clays. This paper anisotropically to the best estimate of the in situ focuses on the so-called lean Drammen clay (Ip = 10- effective stresses and then sheared in compression 17%) which is encountered just below the plastic (CAUC), ranges from 20 kPa at $ m depth to 80 kPa clay. Cross reference to experience from other at 25 m. Sensitivity, as measured by the fall cone Norwegian clays is made when appropriate. test, ranges from 7 to 14 in the same depth interval Sampling, laboratory testing, and analysis with the Fundamental properties of the Drammen clay are aim of evaluating sample disturbance effects are given by Bjerrum (1967) described in this paper. 3 SAMPLING METHODS: 2 DESCRIPTION OF LIERSTRANDA TEST SITE ‘At Lierstranda, the $4 mm NGI piston sampler was Figure 1 shows the location of the Lierstranda test used in addition to the 7S m Japanese piston sampler site just outside the city of Drammen, 35 km south of and the Sherbrooke block sampler. Some other 4 Digitalizado com CamScanner Fig, 1 Location of Lierstranda test site outside Drammen E! soit ‘Water ‘Unit weight |Cone resistance] Undrained shear Z| cecetpton| cont | VINine | ag MPa | stength sy ra [St 10203040 | 18 051015 | 20 40 60 80 TERA 1 ear a cay . 5 (eaesore | 3) er TR] po i fe, * al rn + exam] | ey | to] | iy ssloreor Secu |_| issue a em tbo] | | safe pel | fe = b i 20) |} AR LT le 24 = ani lo. a ale sonar 0 Direct simple shear I Plastic inde, bp + Total unit weigh 1 © CAV compression test Fig. 2 Soil profile from Lierstranda Digitalizado com CamScanner Norwegian sites are referred to in thi the NGI 95 mm piston sampler was eed eens 4 description of this latter sampler is also i seston pler is also included in 41 Sherbrooke block sampler This block sampler was developed and tested at the University of Sherbrooke, Quebec, in the period 1975-1978 (Lefebvre and Poulin, 1979). This special sampler allows the carving of a cylindrical block of diameter 250 mm and height 350 mm at depth from the surface. Cutting procedures similar to those employed for traditional block sampling in shallow trenches are used. Figure 3 shows a schematic view of the sampler with three cutting tools which make an annular slot about 50 mm wide. At each cutting tool, water or bentonite mud is fed from the surface to help removing the cuttings while the sampler is rotated. First a soil sample of about 350 mm in height is carved out. Then components of a horizontal diaphragm, fixed at each cutting tool and contained in the annular slot when in the open position, gradually close as the apparatus continues to rotate. Closure of the diaphragm separates the sample from the deposit and forms a support under the sample when lifting it up in the borehole (see Lefebvre and Poulin, 1979 for more details). Figure 3 gives a schematic view of how a block sample is carved with the support equipment used by Sherbrooke University (Lefebvre and Poulin, 1979). NGI uses a drill rig to operate the block sampler. NGI first had the sampler on loan in 1983. A research programme comparing results of oedometer, triaxial and direct simple shear test results on block samples and on 95 mm piston samples showed the superior quality of the former (Lacasse , 1985). ‘The difference was largest for sensitive, low plasticity clays which are more influenced by sample disturbance than clays of higher plasticity. NGI also had the sampler on loan in 1986 for a research project on Drammen clay (Berre, 1987). In 1994, NGI bought the sampler for use on projects requiring high quality samples. Dimensions and features of the sampler ‘summarised in Table 1 3.2. NGI $4 mm fixed piston sampler (composite version) This sampler was developed and designed by NGI ‘and it is the most common sampler used in Norway. ‘The sampler is a composite piston sampler using plastic inner cylinders (see Fig 4) to prevent corrosion and to avoid practical difficulties with the production and use of stee! cylinders. Preaugering through the dry crust is usually done before starting the sampling. Beneath the dry crust, the displacement method is used; wherein the sampler (with the piston in front of the sample tube) is pushed down to the desired sampling depth without preboring. During sampling the inner rods and piston are fixed in a locked position and the outer rods are pushed down at a constant rate. After withdrawal of the sampler, the sample is sealed at the top by not removing the piston when the cylinder is disconnected from the sampler. For a more detailed description, see Andresen (1981). Dimensions and features of the sampler are summarised in Table 1. 3.3 NGI95 mm piston sampler This sampler was developed and designed by NGI for use in both soft and stiff clays. A tower is used during sampling in order to minimize the number of connections and disconnections. The sampling tube can be connected and disconnected to the sampler when the sampler is hanging from the tower. Fig. 5 shows a longitudinal section of the sampler. The tube is made of cold-drawn mild steel tubes, with a polished inside surface. As for the 54 mm piston sampler, the displacement method is used for lowering the sampler down to the sampling depth. During penetration the piston is kept in its lower position by locking the inner rods and pushing the outer rods at a constant rate. To prevent problems with vacuum at the bottom of the sampling. tube during withdrawal, in some cases a thin plastic tube is fixed to the outer wall of the sampling tube bringing water from the surface down to the cutting edge. For more details, see Andresen (1981). Dimensions and features of the sampler are summarised in Table 1 3.4 Japanese 75 mm fixed piston sampler This sampler has been proposed as the standard in Japan by the Japanese Committee of Soil Sampling Fig. 6 shows the fixed piston, thin-walled sampler as specified by the Japanese Sampling Manual. The boring rod is connected to the upper end of the sampler head and the sampling tube is connected to the lower end (see Fig. 6a). A locking mechanism in the head allows only the upward movement of the Digitalizado com CamScanner Fig. 3 Sherbrooke Block Sampler (ftom Lefebvre and Paulin, 1979) =| = = ‘— vom — ft | Fig.4 NGIS4:mm Andresen, 1981) a4 TT) compare Piston sampler (from Fig. 5 NGL 9S mm piston ‘sampler (from Andresen, 1981) Digitalizado com CamScanner piston rod, thus preventing loss of the sample when the sampler is withdrawn from a borehole. A ball. cone clamp, as shown in Fig, 6b, is typically used as a locking mechanism, Before each sample is taken the standard Japanese ‘method of preaugering down to the sampling depth is used. Special care is exercised to make sure the bottom of the borehole is clean and straight. When sampling, the boring rod and the piston extension rod are connected whl the sampler is lowered smoothly to the bottom of the borehole. During sampling, which is done by advancing the boring rod from the ground surface, the piston is held stationary by the piston rod. After withdrawal of the sampler, the sampling tube is disconnected from the sampler by releasing an air vent on the piston (Tanaka et al., 1996) Dimensions and features of the sampler are summarised in Table 1 4 LABORATORY TESTING METHODS USED ATNGI ‘The procedures for the tests included in this paper are described below. 4.1 Anisotropically consolidated undrained triaxial compression tests (CAUC-tests) The procedure used for CAUC-tests is described by Berre (1982) and consists of the following main steps: The specimen, with a diameter equal to 54, 71 or 80 mm and a height/diameter ratio equal to 1.85 to 2, is mounted into the triaxial cell without allowing any swelling. Then a back pressure is applied that is high enough to give a B-value of at least 95% (i.e. when measured just before start of shearing), The in situ value of the earth pressure coefficient at rest, Ke, is estimated from overconsolidation ratio, OCR, and plasticity index, Jy, using the correlation of Brooker and Ireland (1965). The specimen is consolidated anisotropically to the in situ effective stresses, oy! and Kyo’, these stresses being kept constant for about 24 hours before start of shearing. For specimens with a large degree of disturbance the final consolidation stresses may be kept constant for two days, the criteria being that the rate of volumetric. strain shall be less than 0.0001 % per minute before start of shearin ‘The specimen is compressed (sheared) undrained at an axial rate of strain of 0.55 to 0.75% per hour. During shearing the total radial stress is kept constant while the total axial stress is increased until the specimen fails. Table 1. Summary of dimensions and features of samplers ‘Thickness Area Inside Sampler Length | Dy Dy D, | ratio | clearance ‘Sampling mm mm mm mm % % technique $4 mm piston (NGI) 800 Y i (95 mm piston (NGI) 1000 | 963 | 1016 | 966 | 1 03 Displacement [75 mm piston (Japanese) | 800 15 7 75 | 40 0.0 Preauger Sherbrooke 350 | 250 : : = : Preauger Legend | - . . nae S52 wens este raven + Mig «mr ante ae Lao] Dette cant, ie dare anaes at ore, 1, L Digitalizado com CamScanner 42 Ocdometer tests (Incremental Loading, IL, and Constant Rate of Strain, CRS) The procedures used for IL and CRS oedometer tests are described by Sandbsekken et al, (1986) and consists of the following main steps * The specimen, with a diameter equal to 50 oF ‘67mm and height 20 mm, is mounted into the coedometer cell without allowing it to swell, The specimen is loaded, incrementally (IL) or with ‘constant rate of strain (CRS) to a stress which is ‘equal to about 9 times the preconsolidation stress. An unloading - reloading loop is often done when the vertical stress has become equal to about twice the preconsolidation stress. ‘The purpose of this loop is to be able to correct for sample disturbance for stresses lower than the preconsolidation stress. The time for each load increment for IL-tests is usually 2.5 hours and the rate of strain for CRS-tests is in the range 0.5 to 1.0% per hour. + For CRS-tests, the vertical stress is kept constant overnight at a stress equal to about 2 times the preconsolidation stress to get a point close to the 24 hours virgin compression line. For IL-tests, the stress is kept constant ovemight at about every third load increment. If it is decided to ‘measure the coefficient of permeability, k, directly by constant head tests, this is always done after the stress has been kept constant overnight. ‘The coefficient of consolidation, c,, is calculated from the equation ce, = Mk o where M is the constrained modulus and k the coefficient of permeability. The k-values are checked by constant head permeability tests if there is any doubt about K-values back-calculated from time compression curves (IL-tests) or from measured pore pressure (CRS-tests). The uncertainty in deriving c, from time-compression curves or measured pore pressures directly is then avoided and the e-values can be based on M-values corrected for sample disturbance and k-values at in situ porosity. ‘The ke values are sometimes corrected for the temperature difference between the laboratory and in sity 43 Determination of shear wave velocity ‘The technique of shear wave velocity det makes use of piezoceramic bender elements at ea, end of a soil specimen. The bender element a nd of the specimen is used to generate a shear pulse which propagates axially along the spec The other element is used to determine the arrya, time of the shear wave at the other end of the specimen. Reference is made to Dyvik and M (1985) and Dyvik and Olsen (1989) for more deta The travel time and the specimen height give the shear wave velocity, V,. The shear modulus at smal strain, Gmac, is computed from the formula: (nn @ where p = density of soil. Bender element tests are run in connection with triaxial and oedometer tests. 5 LABORATORY TESTING METHODS USED AT THE PORT AND HARBOUR INSTITUTE (PHRI) IN JAPAN 5.1, Unconfined compression tests (UCT) The UC-tests were done according to the Japanese standard, with a specimen height and diameter equal to 80 and 35 mm, respectively. The axial strain rte was 1%/min, 5.2 Isotropically and anisotropically consolidated undrained triaxial compression tests (CIUC and CAUC-tests) Two kinds of triaxial tests were carried out: iso- ant anisotropically consolidated undrained compression tests (CIUC and CAUC). The specimen size is the same as that of the UCT: the diameter is 35 mm and the height is 80 mm, ‘The consolidation stress CIUC tests is equal to the mean in situ effective stress that is 2/3 of oe (assuming Ky in the Beld being 0.5). In the case of the CAUC tests, assuming that K, is equal to 0.5, the confining and the & stress were increased in steps to the in situ vss (the final vertical stress being equal to o') TH consolidation time for CIUC and CAUC at the Consolidation stage was 2 hours. Axial strain ™# during shearing was 6% per hour. Digitalizado com CamScanner Seren ae (© Danco sting (0) Along scion te 5.00 Pecan Fig. 6 Japanese 75 mm fixed piston sampler (from Tanaka, 1996) 100 Unconfined Compression Test Drammen clay (Lierstranda test side) depth = 12m o 2 4 ee strain (%) Fig. 7 Result of UCTs from 75 mm piston and Block Sample at 12.3 m depth Undrained shear strength, sy, KPa oe 8 . 20] eo 75mm samples, UCT at PHRI ‘75mm samples, UCT at NGI 2s Fig. 8 Undrained shear strength from UCTs vs depth 40 = Block sample & = 4mm sample B 20 5 ‘T5mm saree é 10 ° ° 4 8 2 16 Axial strain, % Fig. 9 Shear stress vs axial strain for CAUC tests at 6,1 m depth 87 Digitalizado com CamScanner 5.3 Ocdometer tests (Constant Rate of Strain, CRS) Constant Rate of Strain (CRS) oedometer tests were performed with a rate of 0.6% per hour. The size of specimen is 60 mm in diameter and 20 mm in initial height. 5.4 Determination of shear wane velocity The shear velocity was measured using bender elements. The soil specimen was placed in an cedometer ring of 60 mm diameter, The initial hheight of the specimen was about 30 mm. The vertical load was applied to the specimen step by step in the same manner as for conventional oedometer tests with measurement of displacement and shear wave velocity. The end of primary consolidation was defined by Casagrande's method. 6 LABORATORY TEST RESULTS FROM LIERSTRANDA TEST SITE 61 UCT In order to make sure that the same procedures were used, the unconfined compression tests carried out both at NGI and at PHRI were performed by PHRI Personnel (Tanaka, 1996). Figure 7 shows comparative results from UCT on Block samp) 15 mn samples at 12 m depth. Figure § ‘home Pe peak shear stress vs depth for UCT carried gu ge 75 mm samples. It can be observed that there does not seem tg any systematic difference in the results of they carried out at NGT and the tests carved out at Piigy This isa cle ion that any sample distu caused by transportation is small compared to te mechanical disturbance caused by tube sampling, 6.2 CAUC tests Figures 9 and 10 show shear stress vs axial strain for CAUC triaxial tests carried out on NGI $4 mm ang 75 mm piston samples, and on Sherbrooke Block samples at depths of 6.1 m and 12.3 m. Figures 1] and 12 show the stress paths from the same tests. It ‘can clearly be seen that the Block samples are of significantly higher quality than the 75. mm piston samples which in tum are of better quality than the 54mm piston samples. Table 2 summarizes the most important parameters from the CAUC tests carried out at NGL. The table includes results from classification tests and measurements from the consolidation and shearing phases, Table 2. Summary of CAUC tests at NGI archange in sample volume ‘divided by inital volume ‘eaxial rat fate; intel vol aio (nat at ous) Digitalizado com CamScanner $0 75mm sample ‘S4mm sample Block sample o 4 8 Axial strain, % R 16 Fig. 10 Shear stress vs axial strain for CAUC tests, at 12.3 m depth * Block samp 8 = em sample q Be ze ‘4mm sample 20 4% 40 (60 60 Effective mean stress, kPa Fig. 11 Stress paths for CAUC tests at 6.1 m Shear stress, kPa 10 20 30 40 50 60 70 2 4000 Effective mean stress, 20 kPa Fig. 12. Stress paths for CAUC tests at 12.3 m 10 Depth, (m) 6 Fig. 13 i strength, sy, KPa 4050 60 70 ‘Block samples, CAUG ‘S4mm samples, CAUC ‘75mm samples, CAUC. ‘75mm samples, CAUC at PHRI ‘75mm samples, CIUC at PHRI ‘Undrained shear strength from CAUC tests vs depth Digitalizado com CamScanner Figure 13 summarizes undrained shear strength vs depth of CAUC tests on 75 mm piston samples as carried out by PHRI and NGI. The tests carried out at PHRI clearly shows higher peak shear stress compared to the tests carried out at NGI. This is discussed later in this paper. 6.3. CRS oedometer tests Figures 14 and 15 show axial strain vs log effective vertical stress for CRS tests on $4 mm and 75 mm piston as well as block samples from 12.3 and 16.4 m depth, ‘Similarly Figures 16 and 17 show enlarged plots of the constrained modulus, M(= s, and coefficient of consolidaton vs stress up to ,' = 500 kPa. Table 3 summarizes the results of the CRS tests including results of soil index tests. and reconsolidation stress, p.' as interpreted using the Casagrande (1936) method. Figure 18 compares results of CRS tests carried out at PHRI and NGT in terms of p.’ as determined by Casagrande's method. 64 Bender element tests Figure 19 shows the computed maximum shear modulus, Gmax =P * Ve, where V, = shear ro velocity and p= soil density vs depth, Measuremen, from both PHRI and NGI are included. NGI hus carried out tests on block samples and 54 mm piston samples during both CAUC tests and CRS tests, IRI carried out tests on CRS samples. A G_. profile as interpreted from Rayleigh wore measurements is also included in Figure 19, 7 EVALUATION OF LABORATORY Test RESULTS FROM LIERSTRANDA In this chapter the results from laboratory test resus are discussed under the following headings: — How sample disturbance influences the different geotechnical measurements described above. = Comparison of results of tests carried out at NGI and PHRI. — Recommendation for quantification of sample disturbance. — Correction of results in order to reduce effects of sample disturbance, — New basis of empirical correlations for interpretation of cone penetration tess. Table 3 Summary of CRS tests at NGI (COEFFICIENT oF |PRECONSO My and cyr=average at Gn Gw=coatficient of 38 In FANGE Oe 10 py Digitalizado com CamScanner Fig. 14 Axial strain, % ‘Axial strain vs log effective vertical stress for CRS at 12.3 m depth Effective Axial Stress, kPa Effective Axial Stress, kPa o 10 200 300000 g g z 3 = * é& z = °o 10 200 at0 ato 0 8 = ig. 16 Constrained modulus and coefficient of consolidation vs effective vertical stress at 12.3 mdepth Effective Axial Stress, kPa 16 |S Axial strain vs log effective vertical stress for CRS at 16.4 m depth seco tom tn to ato 50 2? £ g Ze a4 2, ‘54mm sample 5 # + Tatloading for, My € 9) sen 3 \Block sample & 4 {| roam gave as 3 ema -_ a Fig. 17 Constrained modulus and coefficient of Consolidation vs effective vertical stress at 16.4 m depth Digitalizado com CamScanner Preconsolidation stress, po) KPa Volumetric strain at oye! % o% 120160200240 Se ey Block sampleg © SA samples s 5 wee | 7 75mm saming = ‘ -” . = ‘| € § : i os Ow Block sample, @t NGI 20 20 475mm sample, at NGI + CRSC, by PHRI . . + _ Semen sample, at NGI 2 2 oa Fig. 18. p.'vs depth Fig. 20 Volumetric strain, ee, in CAUC triaxial tests vs depth Gmax» MPa Volumetric strain at a, % oo 1 2 3. 4 5 © Block vax © Block oedometer sh © Shnm ix 7 5mm oedometr at PHRY S 0 = whe & E 4 In stu Rayeigh es é woe tre Bs dj o 6 ae ew . 20 20 pple ct a Fig. 19 Guu fom bender element tests vs depth Fig. 21 Volumetric strain, ein CRS oedometer tests vs depth — Digitalizado com CamScanner 7.1 Signature of sample disturbance on laboratory ‘measurements Volume change during reconsolidation to in situ stresses For many years NGI has used the volumetric strain, Ge, When consolidating back to in situ stresses as an indicator of sample disturbance (e.g, Andresen and Kolstad, 1979). Fig. 20 show ty, as measured in the consolidation phase of the CAUC tests, as a function of depth. The following conclusions can be drawn: + For the block samples there is a tendency for tre to increase with depth from 1.3% at 12 m depth 103.1% at 22 m depth. The & values for the 75 mm samples are clearly higher than for the block samples and quite consistent with depth; 2.1% at 6 m to 2.5% at 16 m depth ‘© The & values measured with the 54 mm samples are consistently higher than the ey values measured by the 75 mm samples; 2.8% at 6 m to 3.9% at 16 m. ‘The same trend was found from the oedometer tests (Figure 21). When compared to the Ew values from triaxial tests, the e. values from the oedometer tests are slightly higher (on the average 0.5% strain). This difference probably results from the different ways in which the triaxial and oedometer samples are prepared. The triaxial samples are prepared by trimming the sample by a thin wire saw. The ‘cedometer samples are prepared by pushing the ste! ‘cedometer ring into the sample, thus potentially causing more disturbance. Hight et al. (1992) came to the same conclusion studying the results of triaxial tests on Bothkennar clay samples trimmed in different ways Peak shear stress in CAUC tests Figure 13 presented the undrained shear strength, 5. (peak shear stress) vs depth. Figure 22 shows the normalized sy-values, s/ow, vs depth. Again the pattern is quite clear: © The tests on block samples give consistantly higher normalized sy-values than the 75 mm tests which again are higher than the tests on 54 mm samples (although not so consistent). * Jou! decreases with depth which reflects that OCR is decreasing with depth. Increasing disturbance with depth may also play a certain role, Strain at failure in CAUC tests Figure 23 confirms the well known experience that axial strain at failure increases with sample disturbance, ‘The following observations can be made: ‘er for all block samples are within a narrow band. © 6 for 75 mm samples are somewhat higher than for block samples, but on the average lower than for the 54 mm samples. er decreases with depth which is believed to mainly reflect decreasing OCR with depth Another contributing factor can be that the plasticity index is decreasing with depth (from I = 20% at 6 m to I, = 14% at 22 m depth). Generally the lower the plasticity the more brittle the clay willbe for a certain OCR-value. The pore pressure parameter D measured in CAUC Inspection of the stress path plots shown in Figures 11 and 12 reveals clearly that the initial part of the stress path before reaching the failure envelope is dramatically influenced by sample disturbance. One ‘way to quantify the effect of sample disturbance on the stress path plots is to use the pore pressure dilatancy parameter, D, which is defined by the following equation: Au = Ada - D (Ac; - Aos) @) where Au = pore pressure change bow= 3 (Aoi +2403) For a CAUC triaxial test (where Ao, = 0): aw l(se ot D [42 4] @) where Ap! = a(sts) and w a(0589 ‘These equations show that D = 0 (ie, zero dilatancy) corresponds to a line wth inclination of 1 horizontal Digitalizado com CamScanner Normalized undrained shear strength, Sy/ovo' Pore pressure param 00 08 04 03 02 oy 920035 040 04s 080 085 000 o—— a) ° 5 8 . 10 ite . a = & . S . v g é, ai So Block samples » 7°11 6 sam samples . 7 75mm samples 2 25 Fig. 22. S/ou'vs depth from CAUC triaxial tets__Fig. 24. Pore pressure parameter D vs depth Axial strain at failure, 7, % oo 2 8g“ ° 20 19] 5 «- + 10%] 16 1 Z| =o . z Bo € . . . - q 8 8 J “ 4 2 20) d 00 os 08 os . Pore pressure ratio B, wt —1 1, som | Fig 25 Cone factor Nu vs CPTU pore pressure Fig. 23 Strain at failure, eg, vs depth Parameter By Digitalizado com CamScanner to 3 vertical on a stress path plot of shear stress vs effective mean stress, provided the same scale is used along the t and p' axes. The traditional pore pressure ameter A for a triaxial compression testis. related toDas follows: A= 1/3 -D, The value of D depends on the stress level; in this study, D has been determined at 2/3 of peak shear stress minus initial shear stress. For the tests at 6.1 rm depth at Lierstranda (Fig. 11) D decreases ftom - 0.05 for a block sample to -0.18 for a 75 mm sample and to -04 for a $4 mm sample. For a perfect specimen one would (except when close to failure) ‘expect a minimum of slippage between particles and therefore a D-value close to zero. In Figure 24, D has been plotted vs. depth Especially at 6.1 and 12.3 m depth at Lierstranda a trend is very clear, with D close to zero for block samples, demonstrating an almost isotropic elastic behaviour, and decreasing with increasing sample disturbance. At 16.4 m depth at Lierstranda the trend is the same, but not so pronounced. Preconsolidation stress, constrained modulus and coefficient of consolidation from cedometer test results Figure 18 shows values of the preconsolidation stress (interpreted according to Casagrande's method) vs depth. Generally the pe’ values from piston samples are lower than from block samples (on the average about 9%). However, the trend is not so clear as for the parameters discussed above. It is worth mentioning that pe! as interpreted from laboratory tests is influenced by the rate of strain. This effect has been reported by several investigators including Leroueil et al. (1983). Some previous tests at NGI on Lierstranda clay samples with 24 h load steps indicated p-values 15-20% lower than obtained from the CRS oedometer tests. The constrained modulus, M, at Lierstranda in the range ow to p.' is lower for piston samples than for block samples (eg. Figures 16 and 17). The difference is in some cases as large as 200%. Above ‘i the modulus is larger for the piston samples than for the block samples. ‘The M-values for the 75 mm Piston samples comes closer to the values for the block samples than the $4 mm samples. At a depth ‘of 12 m the strain when increasing the stress from Ox! to a "common average p.-value’ from $4 mm ‘samples is about 1.5 times that for block samples, and at a depth of 16 m this ratio is about 2 (Figures 14 and 15) ‘The coefficient of consolidation, ¢, at Lierstranda in the range 0! to ps is lower for piston samples than for block samples (e.g. Fig, 16). The difference is in most cases even larger than for the constrained modulus, Small strain shear modulus, Gman measured by bender element tests ‘As mentioned in the previous section, shear wave velocities were measured for some of the CAUC triaxial tests and for some of the CRSC oedometer tests, see Fig, 19, From Fig. 19 it appears that Gnsx measured in connection with CAUC tests are about the same for block and for $4 mm samples. However, the CRSC Gnax measurements give consistently slightly lower values compared to the CAUC tests. This is contradictory to findings reported by Dyvik and Olsen (1989). Fig. 19 also includes Gu values interpreted from Rayleigh wave measurements in the field (BRE, 1990). The Gnax values from the field tests at Lierstranda are significantly higher than the laboratory values. It is believed that this difference is not associated with sample quality, but rather with differences in measurement and interpretation procedures for the two types of tests. Its the opinion of the authors that Gass may not be 4 good indicator of sample disturbance, at least not based on the tests carried out at Lierstranda. 7.2 Comparison of test results from PHRI and NGI Parts of some block samples and most of the 75 mm piston samples were airfreighted from Oslo to PHRI in Japan, Any difference in the test results from NGI ‘and PHRI were then thought to be potentially caused by: () transportation; (ji) length of time between sampling and testing and (ji) difference in testing procedures. As mentioned above, the unconfined compression tests at both NGI and PHRI were carried out by the same personnel from PHRI following identical procedures (Tanaka, 1996) Figure 8 reveals that there is no systematic difference in the results of tests carried out at PHRI and NGL From this observation it is concluded that any potential effects of the three items listed above are minor compared to the disturbance caused by tube sampling From Figure 13 it can, however, be observed that the svalues from the CAUC tests carried out at PHRI are on the average about 20% higher than the ‘corresponding tests carried out at NGI. From the description of testing procedures as outlined in Sections $ and 6 above it is clear that there are important differences in how the testing is performed Digitalizado com CamScanner at PHRI and at NGI. The most significant is very likely the rate of undrained shear which at PHRI is 10 times higher than at NGI, From previous work ‘on Drammen clay and other clays it is a general experience that a 10 fold increase in rate of shearing gives an increase in s, of 10-15%. The results of the Present tests show therefore somewhat higher Aifferencies than previous experience on rate effect has shown, Another difference in procedure is the size of the triaxial specimen which is much smaller at PHRI (dia. = 36 mm) than at NGI (dia, = 71 mm). For the Drammen clay this effect is thought not to be 00 significant, even if no systematic study has previously been done to investigate this. Figure 18 shows the preconsolidation stress, py as interpreted from the CRSC tests carried out at PHRI and NGL. No pronounced differences in test results, are seen, Both PHRI and NGI have used the interpretation procedure proposed by Casagrande (1936) 7.3 Updated criteria for disturbance evaluating sample The influence of sample disturbance on laboratory ‘measurements was discussed above in detail. These results formed the main basis for proposing a new criteria for quantifying sample disturbance. Okumura (1971) listed some quantitative requirements for parameters to be used in evaluating sampling disturbance. Such parameters must be: Easy to determine for the perfectly undisturbed conditions, ji) Regularly variable with disturbance, regardless of the depth of extraction, the stress system experienced, and the soil type. iii) Sensitive to change due to disturbance. iv) Easily and accurately measured, Even though a number of the measure discussed in the previous section are signicanly influenced by sample disturbance (€.. 8c, p My is not possible to satisfy (i) since it will be difficult to know what the measurement should py for an ideal non-disturbed sample. It appears that measurement of Ae/ey OF ex will be the most practical for quantification of samp disturbance because of all the parameters investigated they most consistently satisfy Okumaras criteria (i) and (ii) and best satisty (iv). Also for a perfect specimen Ae/e, Or ey should be close to zero, ‘Ac/es is equal to change in pore volume divided by initial pore volume while ey is equal to change in pore volume divided by initial total volume. Itis reasonable to assume that a certain change in pore volume will be more detrimental to the particle skeleton the lower the initial pore volume. It is therefore suggested to use Aele, rather than 5 when quantifying sample disturbance. Based on the results found in this study and other NG experience the following criteria for evaluating sample disturbance are proposed in Table 4. 7.4 Correction of results fo account for sample disturbance ‘The need to correct laboratory test results in order to account for the effects of sample disturbance has been frequently addressed in the literature (eg Schmertmann, 1955 and Sandbekken et al., 1986). ‘The SHANSEP approach introduced by Ladd and Foott (1974) is also meant to take the effects of sample disturbance into account. The recent research project at NGI has confirmed that it is not possible to find a reliable value of the Preconsolidation stress, ps’, from an oedometer test on a disturbed sample (NGI, 1996). However, ifr’ can be determined by other means, from in situ tests ‘Table 4. Proposed criteria for evaluation of sample disturbance ‘ele, Overconsolidation ratio ‘Very good to excellent <0.04 ‘Good 10 fair (0.04-0.07 Poor Very poor 0.07-0.14 <0.03 (0.03-0.05 0.05-0.10 20.10 For a particular lay multiply Ae/e, by eJ/(1+e,) to get the criteria in terms of toe 4 Digitalizado com CamScanner ‘or geological evidence, procedures have been presented for correcting the oedometer test results to obtain more reliable estimates of constrained modulus, coefficient of consolidation and coefficient of permeability (e.g, Sandbakken et al, 1986 and Karlsrud, 1991), ALNGI work is now underway to update or develop new procedures for camying out and inerpreting twiaxial tests s0 as to correct for the effects of sample disturbance. 7.5 New basis for empirical correlations for interpretation of cone penetration tests: In practice the interpretation of the cone penetration test (CPT) and piezocone test (CPTU) is based on empirical correlations (e.g. Lunne et al, 1989). Most used is the correlation to undrained shear strength, sy 6) eis the corrected cone resistance (eg. Lunne et al, 1989) Gwe is total overburden stress [Nu is the empirical cone factor Obviously the correlation is depending on the quality of the measurement of s, in the laboratory. NGI has chosen CAUC tests carried out on high ‘quality block samples to give the most consistent and reliable basis for the Nu correlations. Now that block sampling (with subsequent laboratory testing) and piezocone testing have been performed in several clays in Norway a new generation of CPTU correlations are underway (Karisrud et al, 1996). Fig. 25 shows as an example Nu vs the piezocone pore pressure parameter Au B, 4 ~ S ‘which has been found to correlate reasonably well 10 OCR (eg Lunne et al, 1989). Au = excess penetration pore pressure measured behind the cone. The range from the data by Lunne et al, (1985) based on CAUC triaxial tests on ‘standard’ piston samples - both onshore and offshore, are also included. It can be observed from Figure 25 that When comparing the new and the ‘old’ data: 1) There is less scatter in the new data, 2) ‘The new Nur values are lower than the ‘old. Use of @ new correlation will therefore result in higher sy-values © Work continues at NGI to improve the CPTU correlations based on high quality block samples 7.6 Use of high quality block sample results in foundation design Present bearing capacity and slope stability analysis are based on many years experience with results from piston samples. The design process involves input soil design parameters, calculation methods and required safety factor(s). Before uncritically using the high quality undrained shear strength parameters in design it is necessary to review the methods of calculation and the required factor of safety. Work is underway at NGI to address this issue. 8 EXPERIENCE FROM OTHER STUDIES ON NORWEGIAN CLAYS ‘The Sherbrooke block sampler has been used at a number of other sites in Norway where piston samples also have been available. A summary of comparative laboratory testing on these samples will bbe given below in order to supplement the findings at Lierstranda. 8.1 Emmerstad, Ellingsrud and Onsoy sites Lacasse et al. (1985) reported tests at two low plastic quick clay sites, Emmerstad and Ellingsrud and one ‘plastic! medium sensitive clay from Onsoy. Table 5 summarizes some of the index characteristics of these 3 clays This study included the following comparative laboratory tests on block samples and NGI 95 mm samples: = CRS oedometer tests - CAUC and CAUE (sheared in extension) triaxial tests ~ direct simple shear (DSS) tests Examples of the CRS oedometer and triaxial test results are shown in Figure 26. For the other test results the reader is referred to Lacasse et al. (1985). Lacasse et al, concluded that the two low plastic quick clays (Emmerstad and Ellingsrud) were significantly affected by sample disturbance, giving similar signatures in laboratory results as the tests on Lierstranda clay detailed in this paper. o7 Digitalizado com CamScanner EFFECTIVE AXIAL STRESS, log 05 (KPa) « ARIAL STRAIN (%) — nox en se a 2) Emmerstad >) Ellingsrud ) Onséy om T SK SF os Y 7 2 Nara sreaw 1%) SHEAR STRESS, (y= 01/2, (hPa) ” onPa m 2 Ennerstad 2) ingsrud ©) Onsiy Fig. 26 Results of CRSC (a) and triaxial (b) tests on block and 95 mm piston samples from Emmerstad, Ellingsrud and Onsoy sites (from Lacasse et al., 1985), i — Digitalizado com CamScanner Table S, Summary of index parameters of three clays studied by Lacasse etal, 1985 Site Emmerstad El Ons Depth sampled (mp 38 anim 310 Water content (1%) 40-48 34-40 58-70 > 60 10 9 o % >I 40 37 60 ‘Liquid li 24-32 25-29 56-74 Plasticity index (@%e) 312 $8 30-44 Sensitivity (FV) 60-20 60-0 69 ‘Overconsolidation Sat4m 2Aat 7m Ta3m ratio. 3at7m 1.4 at 13m L5 at 10m For the more plastic Onsoy clay, the difference in sample disturbance between the block and piston samples was much less pronounced, especially in the CRS and DSS tests. 82 Drammen clay at Elveplassen Berre (1987) reported laboratory test results on block samples, 95 mm piston samples (from Garlanger, 1971) and 54 mm piston samples at Eveplassen in Norway. The tests summarized below were all carried out on low plastic Drammen clay in the depth interval 6-16 m. The vane shear strength varied from about 12 to about 24 kPa, the sensitivity from 10 to 4, natural water content from 31.5 to 39.1%, and plasticity index varied from 11.4 to 15.9%. Berre (1987) concluded that a few of the 95 mm piston samples gave about 14% higher CAUC strength than the $4 mm piston samples. Except for this, the tests on specimens from the piston samples all yielded about the same undrained shear strength values, The tests on the block samples gave undrained compression strength values at least 17% higher than the tests on 54 mm samples. The undrained extension strength was at least 23% higher for the block samples. The extension tests on the block samples had peak shear strength at about -1% axial strain, while the peak shear strength for the 54 mm ‘samples occurred at about -16% strain. 8.3 Leirsund and Fidsvold sites In connection with the railway to the new airport at Gardermoen, 50 km North of Oslo, block sampling was done at some sites to increase reliability of soil design parameters. Karlsrud et al. (1996) described parallel laboratory tests on block and 54 mm piston samples at Leirsund and Eidsvold sites. Table 6 includes some classification data from these sites. The tests at these sites confirm that sample disturbance is very severe for low plasticity clays. Figure 27 compares results of an oedometer test on a block sample and a 54 mm piston sample from the very silty clay at Eidsvold. Figure 28 compares corresponding results of anisotropically consolidated undrained triaxial compression tests (CAUC). The difference between the block samples and the piston samples are again significant, with the following observations: 1x properties for Leirsund and Bidsvold site (from Karlsrud et al, 1996) Table 6, Summary of i [Location | Depth interval | Water content | Wi T | Particles <2 am | x m % % 9 |S20"* Leirsand® T2159 30-39 30-40_| 9-18 Toa Eid $.7-14.7 25-33 31-42 [13-19 378 25 ** one had 240 ‘Approx. 1 km between borings Digitalizado com CamScanner The block sample shows a clearly defined preconsolidation stress, py’ and much smaller strains and higher modulus below p.' than the piston sample. For the piston sample it is also difficult to define, py. At larger strains the block sample is more compressible than the disturbed piston sample. * The CAUC-test on the block sample shows a pronounced peak strength at a very small strain of about & = 0.5%, compared to a large strain of €s = 14% for the piston sample. The fact that the piston sample in this case shows higher peak strength than the block sample is due to a strong ‘tendency for dilation as the effective stress path approaches the Mohr-Coulomb failure envelope. This effect is due to a relatively large volume change of 3.1% during consolidation to in situ stresses for the piston sample, as compared to only 1.0% for the block sample. Thus, the water content at start of sharing is significantly lower for the piston sample. These effects of poorer sample quality may be explained by the tendency for the original clay structure to dominate the behaviour ‘at small stresses and strains, whereas the absolute void ratio dominates the behaviour at larger ‘strains. stresses and the stress strain behaviour in the triaxial tests and oedometer tests showed that essentially all block samples were of extremely good quality, and much better then the 54 mm piston samples. The effects of poorer sample quality on test results depends, however, on the type of clay in question as ‘was concluded by Lacasse et al. (1985) as outlined above 9 SUMMARY AND CONCLUSIONS. ‘The following conclusions are primarily based on laboratory tests on block samples and parallel piston samples from the low plastic Drammen clay taken from NGI's test sites at Lierstranda. They are also consistent with experience obtained from several ‘other Norwegian sites with parallel block and piston samples. 1. Block samples have been successfully obtained to depth of 22 m below ground level using the ‘Sherbrooke sampler. ‘aca all sre, oP) Pe er ee ee) Fig. 27. Results from oedometer tests from Eidsvoll (from Karlsrud et al, 1996), a a a onto, p= Ste aro) Results of CAUC triaxial tests from Eidsvoll (from Kartsrud et al, 1996). 4 Digitalizado com CamScanner

You might also like