In-Situ Evaluation and Calibration of Radioisotope Cones in Singapore Marine Clays

Transcription

1 CHAPTER 3 In-Situ Evaluation and Calibration of Radioisotope Cones in Singapore Marine Clays 3.1 Introduction The working mechanism of a Radioisotope Cone Penetrometer has been described by Shrivastava (1994) and the effectiveness of these cones in naturally deposited clayey and sandy soils in Japan has also been reported by Shibata et al. (1993) and Mimura et al. (1995; 1999). The calibration and evaluation of these cones is one of the important tasks to ensure reliable measurements. In this chapter, the insitu performance of a RI Cone Penetrometer to measure the wet density and water content of soil, which are the two important parameters in characterization of lumpy fill, is evaluated by comparing the data from field tests with laboratory test results carried out on undisturbed samples in a major site investigation in Singapore. Laboratory experiments are also carried out to evaluate the statistical fluctuation of ND-CPT and this will be used to demonstrate the effectiveness of averaging measuring RI Cone data. Also, the applicability of the calibration charts, which were obtained for soils in Japan, to soils elsewhere, has been studied. Finally, an in-depth examination of the accuracy of the RI cone results is also presented. A significant portion of this chapter has been accepted for publication in ASTM Geotechnical Testing Journal (Dasari et al. 2005). 3.2 Description of RI Cone Penetrometers Figure 3.1 shows two different RI Cone Penetrometers, namely a NM-CPT and a ND-CPT. The lower part of the cone houses various sensors to measure the usual 33

2 cone parameters, namely, cone resistance (q c ), pore pressure (u 2 ), and the sleeve friction (f s ). The size of the lower part conforms to the standards recommended by the International Society for Soil Mechanics and Foundation Engineering (ISSMFE, 1989) reference test procedure on cone penetration testing. The diameter of the cone is 35.6 mm and the apex angle is 60. The base area of the cone is 10 cm 2 and the area ratio, (a) is equal to A porous ceramic filter is located just behind the cone tip. The total length of the shaft housing the sensors is 258 mm. After this, the shaft tapers outwardly at an angle of 15. The tapered portion of the shaft is 49 mm long and above which the shaft has a constant diameter of 48.6 mm and extends for a total length of 896 mm. This upper part houses the Radioisotope source, the detector, and a preamplifier Neutron-Moisture Cone Penetrometer (NM-CPT) Neutrons are slowed down in the presence of hydrogen and this characteristic is used in the NM-CPT. If it is assumed that all the hydrogen present in the soil is in the form of water then slowed down neutrons are related to the water or moisture content of the soil. The NM-CPT uses Californium isotope ( 252 Cf) as the fission source of the neutron with a half-life of 2.65 years. The detector used in the cone penetrometer is a Helium (He 3 ) filled proportional tube. Fast neutrons are emitted from the source. After several collisions, the neutrons lose their energy and the detector will capture these slowed neutrons. This will give a measure of the hydrogen in the soil, which in turn is related to water or moisture content of the soil. Figure 3.1(a) shows the schematic diagram of a NM-CPT. 34

3 3.2.2 Nuclear-Density Cone Penetrometer (ND-CPT) The ND-CPT uses gamma-ray as the source. Gamma rays interact with soil predominantly depending on the level of the energy. The Compton scattering is predominant within an energy range of 600 kev and 1.2 MeV, and is a function of the material density. If the detector is so designed that it measures only the incoming photons within the range described, then the incoming photons are a function of the density material only and is given as follows: I ( µ. x) = I exp (3.1) 0 m ρt. where ρ t : wet density; I 0 : incident radian intensity; I : transmitted radian intensity; µ m : total mass absorbing material; and x : thickness of the absorber. The gamma-ray source used in the construction of the ND-CPT is the Cesium isotope ( 137 Cs) with a half-life of 37.6 years, and the detector is a sodium iodide activated with thallium (NaI (TI)) scintillator mounted on the photomultiplier tube. The length of the NaI scintillation detector is 10.2mm. The separation distance between the source and center of gamma detector is 255mm. Figure 3.1(b) shows the schematic diagram of a ND-CPT. Figure 3.2 shows the working principle of density and water content measurement using the RI cones. The density and water content are measured using a source and detector. The signals from the source interact with soil and the modified signals are captured by the detector. The degree of modification of the signals is a function of material property. RI cone measurements (water content and density) are evaluated in an extended volume around the central point of the radioactive source and detector configuration, which is called the measuring volume. Based on the theory of gamma scattering and neutron methods (Homilius and Lorch, 1958; Olgaard, 1965), the maximum sphere of the influence zone around the source was found to be about 35

4 30cm radius and these measurements therefore provide an average density of the material within the measuring volume. To re-examine the maximum radius of the influence zone for ND-CPT, an experimental investigation was also conducted and this will be discussed in the following chapter (Chapter 4). 3.3 Calibration of RI Cone Penetrometers It is important to calibrate the NM-CPT and ND-CPT to obtain an accurate relation between density/water content to the count rate ratio. The count rate ratio is defined as a ratio of the actual nuclear-density (RI) count to the standard count, which is measured in an inactive material under controlled condition. The count rate ratio, rather than the actual nuclear-density (RI) count, is used for calibration because radioactive sources weaken with the passage of time due to decay. The effect of this change in the intensity of sources on in-situ measurement can be corrected by adopting the count rate ratio as the measure for correlating with density or water content. The error in wet density caused by the use of count rate ratio, rather than direct RI count is negligible. Calibrations are usually done in uniform soils in laboratory such as those conducted by Shibata et al. (1992; 1993). RI cones were pushed into sand or clay samples of uniform water content, and density. Calibration charts relating density/equivalent water content to the count ratio were then obtained. The reliability of the laboratory calibrations curves should be checked against field data. However, the field water content and density are unknown and an independent method is necessary to measure them. Shrivastava and Mimura (1998) and Mimura et al. (1999) carried out field calibration studies in clay and sand. The field density and water content were determined in the laboratory from samples retrieved using the thin wall 36

5 sampler for clay and the freezing technique for sand. The field calibration curves of water content and wet density that were obtained using NM-CPT and ND-CPT respectively are shown in Figure 3.3. The field calibration equation for NM-CPT and ND-CPT is given below: R m = ρ m * ρ m * 2 for NM-CPT (3.2) R ρ = ρ t ρ 2 t for ND-CPT (3.3) where R m is the moisture count ratio, ρ m is the equivalent volumetric water content (t/m 3 ), R ρ is the density count ratio and ρ t is the wet density (t/m 3 ). As can be seen in Figure 3.3, three different soils were tested by Nobuyama (2000), and there is a consistent trend between count ratio and equivalent water content and wet density of soil. It is important to note that there is a unique calibration curve for both clay and sand. If this uniqueness also holds true for different locations, then this would greatly simplify the use of this device, and also allow such calibration curves to be universally used. Thus examination of the applicability of the calibration curves, which were obtained for soils in Japan, to soils in Singapore, is a focus of this chapter. Figure 3.3 also shows that the bulk of the data are confined within a narrow ± 5% band, which gives an indication on the limits of accuracy of measurements using the RI cones. The readings in these figures are also affected by various correction factors, which were described earlier in Chapter Procedure for Processing the ND-CPT Results Current design of the ND-CPT requires two probings for every single measuring point, one probing to obtain the background (BG) count of naturally occurring gamma photons and another probing to measure the actual nuclear density (RI) count. The background count is measured using a dummy cone, in which only the 37

6 detector is placed to measure the naturally occurring gamma photons. This natural radioactive (background) count is a type of noise which must be subtracted from the count measured to give the actual nuclear density measurement. To determine the wet density and water content of soil, the measured raw data needs to be processed for the final profiles through the following steps: (i) depth corrections (ii) noise suppression or elimination (iii) statistical fluctuations and averaging of measured RI cone data. Figure 3.4 also shows the flow chart for processing of measured RI cone data. The detailed procedures for processing of measured RI Cone data will be discussed subsequently Depth Corrections As stated earlier, the upper part of the RI cone houses the radioisotope source, the detector and a preamplifier. The lower part of the cone houses various sensors to measure the usual cone parameters, namely cone resistance, pore pressure and the sleeve friction. Therefore, it is important to correct the measurement of the different sensors reading into the same depth level as the other parameters to ensure proper characterization of soil layers. Therefore, the following depth correction needs to be carried out for raw data for further interpretations. If the measurement center for a RI cone is considered at the cone resistance sensor (load cell), then the other sensors readings are corrected accordingly, for example, the depth of sleeve friction, pore pressure, RI count, BG count data s are raised about 0.11m, 0.04m, 0.60m and 0.36m respectively Noise Suppression or Elimination In several practical cases, it is possible that the depth-meter reading could be affected by either delayed response or disturbing inertial effects, caused, for example, 38

7 by velocity fluctuations (acceleration, vibrations, and so forth). Hence, it will affect the accuracy of RI cone measurement. To avoid this discrepancy in the RI Cone results, it is important to measure the RI Count and BG Count in both penetration and uplift stage so that the retardation error can be pointed easily. The analog recording of the density depth profile then will show retardation hysteresis loops, which enclose the representative density profile. Figure 3.5 shows the typical BG count profile measured for both penetration stage and uplift stage profile. From the figure, it can be seen that there is an abnormal behavior observed in the penetration stage measurement due to undesirable shock. These undesirable shocks can be removed by comparing with the uplift stage measurement. Figure 3.5 illustrates the typical BG count profile after elimination of the undesirable noise Averaging of Measured RI Cone Data Radioactive decay is a random process. Consequently, any measurement based on observing the radiation emitted in nuclear decay is subject to some degree of statistical fluctuation. These inherent fluctuations represent an unavoidable source of uncertainty in all nuclear measurements and often can be the predominant source of imprecision or error (Knoll, 2000). The source-originated fluctuations affect the precision of the RI cone. The error in density/water content due to these fluctuations can be minimized by designing equipment such that the radioactive count is high. The error can also be minimized by decreasing the statistical fluctuation through filtering data over a short depth. Nobuyama (2000) reported that the RI Cone measurement needs to be averaged over a 10cm depth in order to reduce the statistical fluctuation. To re-evaluate the statistical fluctuation of RI cone measurement, a laboratory counting experiments is carried out and discussed in the following paragraphs. The 39

8 understanding of these statistical fluctuations is also important in the interpretation of the ND-CPT results for highly heterogeneous soils. 3.5 Statistical Fluctuation of ND-CPT A series of laboratory counting experiment was carried out using the ND-CPT under identical conditions. The calibration chamber used for laboratory counting experiments is shown in Figure 3.6. One of the controlling factors is the size of the laboratory calibration chamber. As stated earlier, based on the theory of gamma scattering and neutron methods, the measuring volume around the source was found to be about 30 cm radius. Therefore, a stainless steel chamber with diameter of 700 mm and height of 1000 mm was used for the experiments. This calibration chamber was filled with water and the ND-CPT was kept inside at the centre of the calibration chamber to measure the statistical fluctuations of RI Count and BG Count under identical condition, as shown in Figure 3.6. Figure 3.7 shows the typical variations of RI count and BG count data measured in water under ideal condition. There are 121 separate measurements, each taken at a 1 sec interval. The source of the radiation was steady ; i.e., during the 121 secs when the measurements were being made, the source of radiation did not change in its nature. Nevertheless, the number of counts recorded per sec is clearly not constant. This is the statistical nature of ND-CPT measurement. The frequency of occurrence of an error of any given magnitude can be calculated for a given determination by the application of the laws of probability. Various distribution laws are used to relate the magnitude of a deviation from true average and the number of events which experiences that particular deviation. The normal distribution is useful for describing the spread in data occurring in various 40

9 aspects of nuclear-radiation detection (Price, 1964; and Tsoulfanidis, 1995). The extent of the statistical fluctuations about the true mean may be expressed in terms σ, the standard deviation and is approximated by normal distribution (Nobuyama, 2000). The standard deviation is defined as the square root of the scatter data about the mean value. The standard deviation is used commonly to indicate the accuracy with which a sample of a given size can estimate the mean value of test results. The coefficient of variation, COV, is another measure of variability of a sample of test results and defined as the ratio between the standard deviation and mean value. The COV expresses the magnitude of variability as a fraction or as a percentage of the mean value. This facilitates comparison of data from different samples because the mean and standard deviation tend to change together, so typically the coefficient of variation remains fairly stable (Snedecor and Cochran, 1980). Figure 3.8 illustrates the form of normal distribution along with its mean and standard deviation of measured RI count and BG count data. These figures show very clearly the amount of fluctuations present in the ND-CPT measurements can be approximated by Normal or Gaussian distribution. It is also noted that the ND-CPT measurements fluctuate within the range of ± (twice of the standard deviation, σ) from its mean value. The coefficient of variation for RI count and BG count data s is 2.4% and 15.1% respectively. Nobuyama (2000) reported that the error could be minimized through filtering the data over a short depth. For this, the moving-average technique has been employed for the measured raw data at different time intervals. Figure 3.9 shows the changes in COV for different moving-averaging interval for RI Count and BG count data s. From the figures, it can be seen that the COV decreases with increasing moving-average time interval. It is also inferred from Figure 3.9 that the COV is fairly constant after a moving-average interval of 6 secs to 7 secs. This further 41

10 confirms that the averaging or filtering of wet density profiles measured by ND-CPT is necessary to reduce the statistical fluctuations. In field test, the ND-CPT was pushed into the ground at a rate of approximately 1.5 cm/sec, and thus it will cover a distance of 10cm within 6 7 secs. While comparing the BG count with RI count in water, the BG count is too small and it is almost equal to the minimum detectable count of ND-CPT. Therefore, if the effect of the BG count is neglected, as can be seen in Figure 3.9 (a) the error due to the statistical fluctuation in ND-CPT measurement is less than 1% if an averaging span of 10 cm is used. This can be also demonstrated using the field results shown in Figure 3.10 (a) and (b) which were obtained by averaging over 5cm and 10cm depths respectively. The raw data are also shown for comparison purpose. As can be seen, when averaged over 10cm depth, the RI cone profile count distribution trend is not distorted or deformed compared to the original raw data, which means the loss of information about the density and moisture through the filtering is negligible. This filtering is necessary to smooth the density profile. Therefore, the RI cone measurements that are reported subsequently are averaged over a span of 10cm depth to minimize the statistical fluctuations. 3.6 Site Investigation for In-Situ Evaluation of RI Cone The geographical location of the test site is shown in Figure The Punggol Timor Island in Singapore was reclaimed using large dredged clay lumps during Large dredged clay lumps of up to about 8m 3 in volume were excavated along the periphery of the island using a clam-shell grab. They were transported using bottomopening barges and dumped onto the island seabed to form a lumpy fill layer about 8m thick. Sand was dumped on top of the lumpy clay fill to compress the clay lumps. Such 42

11 a reclaimed ground is expected to be highly heterogeneous with a complicated mixing of soft clay lumps and sand. Parameters of special importance in the characterization of this type of ground are the density and water content. Radioisotope Cone Penetration tests, undisturbed soil sampling and laboratory tests were carried out to measure wet density and water content. This provides an opportunity to evaluate the performance of the two types of RI cone penetrometers described earlier, and also to characterize the present state of ground. In this chapter, the in-situ performance of RI cone penetrometer is evaluated. The present state of this reclaimed land will be discussed in Chapter RI Cone Penetration Tests The RI cone penetration tests were performed according to the same procedure as for standard piezocone penetration tests specified by British Standard BS 5930 (British Standards Institution, 1999) and the International Society for Soil Mechanics and Foundation Engineering (ISSMFE, 1989). The RI cone penetration test consists of cone penetration test (CPT) and density/water content measurement using radioisotope source. During testing, these cones were pushed into the ground at a rate of approximately 1.5 cm/sec and cone tip resistance (q c ), sleeve friction (f s ), pore pressure (u 2 ), RI count and BG count were recorded continuously. The porous filter elements were inside the water for saturation until they are assembled in the piezocone in the field Boreholes and Undisturbed Samples Boreholes were located as closely as possible (about 0.5m) to RI-CPT locations so as to enable comparison of the results from the cone tests with those of undisturbed 43

12 samples from the boreholes. The need for such proximity arises as a result of the expectation of a highly heterogeneous ground. The largest lumps can be of size of about 8m 3 and typical clay lump sizes vary from 0.5m to 2.0m. Thus, this close spacing is needed to enable a better understanding of the ground. The undisturbed soil samples were obtained using an open drive sampler together with 100mm diameter thin-walled sample tubes. The edges of the tubes were tapered to 6 0 ±1 0 to minimize soil disturbance. The usual length of the sample was 1000mm, but some 2000mm long samples were also collected using longer tubes. The sampling tubes were waxed at the ends immediately after retrieval to prevent the loss of water content Laboratory Testing A laboratory-testing program was undertaken to determine the wet density and water content of soils, these being the most critical parameters to evaluate the performance of the two types of RI-CPTs and also these are considered as primary parameters in the characterization of a reclaimed heterogeneous ground. The soil samples were extracted by mounting the soil sampler horizontally on an extruding machine. The soil samples coming out of the sampler were supported by a halfcylinder to minimize disturbance. As discussed previously, RI cone measurements are averaged over a span of 10cm depth to reduce the statistical fluctuations. Therefore, the soil samples from the sampling tube were cut into pieces of about 10 cm lengths. The weight and volume of each sample were measured to determine the wet density of that sample. The water content was measured using trimmed materials, according to the procedures specified by British Standard BS 1377 (British Standards Institution, 1990). 44

13 In addition, an extensive testing program was undertaken to evaluate the stress history, strength, and compressibility characteristics of the dredged clay layer. Conventional oedometer and undrained and unconsolidated (UU) triaxial tests were conducted on the undisturbed soil samples retrieved from the site. Low (2004), a coreseacher, has proved that the soil samples obtained using the thin-walled sampling tube gives a good quality samples according to the classification proposed by Lunne et al. (1997a). These tests were performed according to the British Standard BS 1377 (British Standards Institution, 1990). Grain-size distribution and Atterberg limits were also performed Soil Profile A typical set of measured cone tip resistance and pore pressure response at the test site is shown in Figure The pore pressure response suggested that the water table was about 3m below ground level and the pore pressure increased linearly up to about 14m with occasional fluctuations. The linear increase in pore pressure follows the hydrostatic line and indicates that sand is present up to 14m. The deviations from linearity at 8m and 12m depths indicate presence of a less permeable layer, later through visual inspection, confirmed to be a sand-clay mix. The sudden increase in the pore pressure below 14m suggests that clay is present below this depth. In the present site investigation, soil samples from sand layer were not retrieved as freezing technique is very expensive and rarely used in Singapore. The clay-sand mix layer below the sand was found to be a highly variable layer even within a distance of 0.5m as the present site was reclaimed using clay lumps. The clay layer below 14m shows a more consistent degree of uniformity, with some pockets of dense 45

14 and soft materials. Only data from this layer were used to evaluate the performance of the Radioisotope cone penetrometers. 3.7 Field Assessment of Radioisotope Cone Penetrometers As radioisotope cone penetration testing is new, the cost is high (over US$1000 per test) and it is prudent that a series of preliminary tests was conducted prior to the investigation proper. Also, the principal use of this cone will be mostly under marine conditions where the soil is expected to be fully saturated. Thus the initial tests will provide an evaluation on whether both the ND-CPT and NM-CPT are needed or not. If the soil is fully saturated, and if the measurements are accurate, then only one test is needed, from which both the wet density and water content can be established Preliminary Site Investigation In the preliminary site investigation, six ND-CPTs, six NM-CPTs and 2 boreholes were conducted, as shown in Figure These tests were spaced very closely to compare results from the cone tests with those from laboratory tests conducted on undisturbed soil samples retrieved from the boreholes. The wet density and water content profiles obtained from these tests are compared with laboratory measurements and the results are shown in Figure 3.14 and The results CPT 1 - CPT 4 are compared with that from their nearest borehole BH-1, similarly CPT 3 CPT 6 are compared with that from BH-2. The wet density and water content results were shown only for depth below 14m, as the clay layer below 14m displayed a consistent degree of uniformity. As can be seen, there is a reasonable agreement between RI-CPT measurements and laboratory data obtained from the undisturbed soil samples retrieved from the boreholes. However, there are some differences in the 46

15 measurement between the RI cone results and laboratory data. This may be attributed to the result of highly heterogeneity in this reclaimed fill. The comparison of water content measured directly by NM-CPT and the computed water content from the density measured by ND-CPT, assuming soil is saturated, is shown in Figure There is a very good agreement between the results from the two different types of RI cones, which implies that any one of the cones could be used for fully saturated soils. Note that there are significant deviations in the first 3m because the soil is not fully saturated. As discussed previously in Chapter 2, the ND-CPT and NM-CPT measurements are affected by several factors. As the ND-CPT has a lesser number of corrections to be applied compared to NM-CPT, for the detailed site investigation, only ND-CPTs were used Detailed Site Investigation The detailed site investigation consists of: (a) 86 ND-CPT (b) 5 standard piezocone penetration tests (PCPT); (c) 12 boreholes and (d) an extensive laboratory testing program including both routine and advanced tests to determine engineering properties of the reclaimed land. The site investigation plan carried out at the Pulau Punggol Timor site is shown in Figure 3.17(a), which consists of a big grid and two small grids. The big grid is 25.5m x 22m and the small grids are 2m x 2m. The big grid contains 79 ND-CPTs, 5 PCPTs and 11 boreholes, as shown in Figure 3.17(b) and the small grids have seven ND-CPTs and 1 borehole. Boreholes were located as closely as possible (about 0.5m) to ND-CPT locations so as to enable comparison of the results from the cone tests with those of undisturbed samples from the boreholes. The present test arrangement was designed to provide adequate information to characterize a representative volume. Therefore, to obtain an accurate assessment of the 47

16 heterogeneity in the lateral extent, an atypically small and practical spacing of 0.5m was selected at the centre of the grid. The spacing was increased towards the edges of the grid. By usual geotechnical standard, this layout is unusually close. However, for this particular project, such a close spacing was needed, as the aim is to characterize a layer formed by placement of clay lumps, whose sizes varied from 0.5m to 2m. The disturbances due to such small spacing were minimised by taking special care in maintaining verticality of both cones and boreholes. 3.8 Comparison of Laboratory and ND-CPT measurements Figures 3.18 and 3.19 shows the comparison of in-situ wet density profiles measured using the ND-CPT together with the laboratory wet density measurements independently obtained from undisturbed samples as well as wet density calculated from the measured natural water content. The results from only five out of the 86 ND- CPTs are shown in these figures. The data from each of the five boreholes are compared with a ND-CPT nearest to it. As can be seen in these two figures, there is good agreement between ND-CPT measurements and laboratory data. Figures 3.18 and 3.19 also show the results obtained when calibration curves ± 5% offset from the best fit curve, as shown in Figure 3.3, were used to calculate the density. Except for a few measurements, most of the measured values lie within these bounds. This suggests that the RI cone used has an accuracy of ± 5% in determining the wet density and water content of soil, a fact already indicated in the calibration chart shown in Figure 3.3. Some of the laboratory wet density measurements are outside the ± 5% bounds. It should be noted that big differences could be noticed when the density is either too high (> 20 kn/m 3 ) or too low (< 16 kn/m 3 ). These measurements are likely to be a 48

17 result of heterogeneity in this reclaimed fill. The lower density is due to presence of organic material and the higher density is due to presence of sandy material. As explained earlier, the cone measures density over a radius of 30cm distance, therefore results from the cone are averaged values rather than spot values. In spite of some differences, it is shown that ND-CPT has performed very well in measuring the wet density. 3.9 Statistical Analyses Statistical analyses were performed between the field ND-CPT measurements and laboratory data. A total number of 103 laboratory data points and the field values at the corresponding depths were chosen for the analyses, and are tabulated in Table 3.1 and 3.2. The laboratory data points and the corresponding field values are also shown in Figure 3.20 (a). Figure 3.20 (b) shows histogram of frequency of data points versus the measurement bias (β), which is the ratio of actual laboratory and ND-CPT measurements. The mean correction factor and coefficient of variation are 1.01 and 5.9% respectively. It can be calculated that 80% of field data points lie within ± 5% from the laboratory measurements. As discussed previously, some of the deviations are a result of the heterogeneity expected in this particular ground that is even at such closing spacing, the two densities are from two different materials. These 13 data points are highlighted in Table 3.1 & 3.2 and shown in Figure 3.20 (a) as solid circles, and were removed for subsequent analysis. The remaining 90 data points are incorporated into Figure 3.3 (b) to get an improved calibration curve for Singapore clays, and shown in Figure 3.21(a). The improved calibration equation for ND-CPT is given below: 49

18 R ρ = ρ t ρ t 2 (3.4) The results from statistical analyses are shown in Figure 3.21(b). As expected, the mean and coefficient of variation improved to 1.0 and 3.0% respectively. With this removal, it can be calculated that about 92% of data points lie within ± 5% from the mean. As shown in Figure 3.3, the calibration data have ± 5% bounds. The measurement error indicates that the ND-CPT measured values are within the limits of accuracy dictated by those inherent in the calibration curves. From these results, it is clear that ND-CPT performed very well in measuring the wet density of soil Concluding Remarks A major site investigation was carried out to evaluate the in-situ performance of the Radioisotope cone penetrometer in a highly heterogeneous clayey ground. Insitu water content and wet density were measured using the NM-CPT and ND-CPT respectively. It was found that for saturated clay layer, the water content calculated by the ND-CPT agreed very well with the water content measured by the NM-CPT. For fully saturated soils, therefore, the use of ND-CPT is recommended as this cone is influenced by a lesser number of factors compared to NM-CPT. Independent data on water content and wet density were obtained from laboratory testing on the samples retrieved using high quality samplers. The comparison of the RI cone data with the laboratory test measurements showed very good agreement. The calibration charts derived for soils in Japan were found to be suitable for Singapore marine clay. The uniqueness of the calibration curve for different soil and locations is an advantage in the use of such cones. An improved calibration chart by incorporating the data from Singapore marine clay is also obtained. About 92% of field density measurements using ND-CPT were found to be within ± 50

19 5% of the laboratory measurements, this range is consistent with the spread inherent in the calibration data. It was also found that the averaging or filtering of wet density profiles measured by ND-CPT is necessary to reduce the statistical fluctuations. Further, the other key aspects such as the influence of measuring volume and the interpretation of signatures of ND-CPT results for layered and heterogeneous lumpy fill need to be investigated. To explore these issues, an experimental investigation will be conducted and discussed in next chapter (Chapter 4). 51

20 Serial No. Borehole No. Depth (m) Lab measured wet density (kn/m 3 ) Density Count ratio RI cone measured wet density (kn/m 3 ) 1 BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH Table 3.1. ND-CPT versus laboratory measured wet density data (Note: highlighted data points are lie outside ± 5% bound). 52

21 Serial No. Borehole No. Depth (m) Lab measured wet Density Count RI cone measured density (kn/m 3 ) ratio wet density (kn/m 3 ) 51 BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH BH Table 3.2. ND-CPT versus laboratory measured wet density data (Note: highlighted data points are lie outside ± 5% bound). 53

22 (a) (b) Figure 3.1. Diagrams of RI cone penetrometers (a) NM-CPT (b) ND-CPT (after Shibata et al., 1992 and 1993). NM Cone ND Cone Sphere of influence zone RI Sources Soil Layer Soil Layer Figure 3.2. Working principle of measurement in RI Cone Penetrometers. 54

23 2.0 R m = ρ m * ρ m 2 (N=278, r 2 = 0.986) -5% Regression Moisture Count Ratio, Rm (a) +5% Regression Kinkai Bay (Clay) Hachirogata (Clay) Higashi-Ogishima (Sand) Water (Laboratory Test) Equivalent Moisture Content, ρ m (t/m 3 ) R ρ = ρ t ρ t 2 (N=354, r 2 = 0.99) Density Count Ratio, R ρ % Regression (b) +5% Regression Kinkai Bay (Clay) 0.5 Hachirogata (Clay) Higashi-Ogishima (Sand) Wet Density, ρ t (t/m 3 ) Figure 3.3. Field Calibration Curves for RI Cone Penetrometers (a) NM-CPT (b) ND-CPT (after Nobuyama, 2000). 55

24 Test Start RI Cone Test Sampling time:1 sec Raw data (Count rate:cps) Depth Correction Noise Suppression Average Count for 10cm depth span Count Rate Ratio Correction for Source Decay Calibration Equation Directly detected item, ρ t, ρ m* ' α ' Correction for water content Derive the ρd, w, & other parameters End Figure 3.4 Flow Chart for Processing the RI Cone Results 56

25 Figure 3.5 An example of Noise Suppression or Elimination (after Nobuyama, 2000) 57

26 ND-CPT Diameter = 700 mm Height = 1000 mm Figure 3.6 Cylindrical Calibration chamber used for laboratory counting experiments. 58

27 RI Count (CPS) Time (Secs) Background Count (CPS) Time (Secs) Figure 3.7 Typical variations of RI Count and BG Count data measured in water under ideal condition. 59

28 0.015 Probabaility Mass Function (a) σ σ σ σ Raw Data Mean = 1404 SD = 34 COV = 2.4% N = RI Count (CPS) 0.07 Probabaility Mass Function (b) Raw Data Mean = 39 SD = 6 COV = 15.1% N = σ σ σ σ BG Count (CPS) Figure 3.8 Illustrating the statistical fluctuations of RI Count and BG Count data in the form of normal distribution along with its mean and standard deviation (σ) (a) RI Count (b) BG Count. 60

29 (a) RI Count Coefficient of Variation (%) Equal to 10cm Span in actual field data Moving average interval in secs (b) BG Count Coefficient of Variation (%) Equal to 10cm Span in actual field data Moving average interval in secs Figure 3.9 Changes in COV with different moving averaging intervals (a) RI Count (b) BG Count. 61

30 RI Count (count per sec) RI Count (count per sec) (a) 4 (b) Depth (m) Depth (m) Raw data 5cm average 18 Raw data 10cm average Figure 3.10 Comparison of measured ND-CPT raw data and averaged data (a) 5cm average span (b) 10cm average span. 62

31 Scale: 1: Figure Geographical location of the test site in Singapore Cone Resistance, qc,(mpa) Pore Pressure, u2, (MPa) Sand Depth (m) 10 Depth (m) 10 Sand + Clay Mix Clay Figure Typical cone resistance and pore pressure profile at test site. 63

32 CPT 1 1.0m 1.0 m CPT 3 CPT m BH 1 BH 2 CPT 2 CPT 4 CPT 6 Figure Layout RI-CPT and boreholes for preliminary site investigation 14 Water Content (%) Water Content (%) BH CPT - 1 CPT BH - 2 CPT - 3 CPT - 3 CPT CPT CPT - 5 CPT - 6 Depth (m) 17 Depth (m) Figure Comparison of measured NM-CPT water content profiles with laboratory results. 64

33 14 Wet Density (kn/m 3 ) Wet Density (kn/m 3 ) CPT - 1 CPT - 2 CPT - 3 CPT - 4 BH CPT - 3 CPT - 4 CPT - 5 CPT - 6 BH 2 Depth (m) 17 Depth (m) Figure Comparison of measured ND-CPT wet density profiles with laboratory results. Water Content (%) CPT 2 2 Water Content (%) CPT ND-CPT NM-CPT 4 6 ND-CPT NM-CPT 8 8 Depth (m) 10 Depth (m ) Figure Comparison of measured NM-CPT water content profiles with water content calculated from ND- CPT. 65

34 50m (a) 2m 36 2m 72 BH Big Grid 25.5m x 22.0m 35 2m 2m m ND-CPT Boreholes (b) 69 BH m BH BH 7 BH BH BH BH BH 5 52 BH BH BH Legend ND-CPT PCPT Boreholes m All dimensions are in metres Figure 3.17 Layout of RI-CPT and Boreholes in Site Investigation (a) Location of small grids (b) Big grid. 66

35 Wet Density, (kn/ m 3 ) Wet Density, (kn/ m 3 ) Wet Density, (kn/ m 3 ) RI 17 BH 7 - direct measurement BH 7 - from water content 15 RI 21 BH 8 - direct measurement 15-5% +5% BH 8 - from water content RI 25 Depth (m) 16 BH 7-5% +5% Depth (m) 16-5% +5% Depth (m) 16 BH 9 - direct measurement BH 9 - from water content 17 RI R BH 8 17 RI m 0.5 m RI 21 BH m 0.5 m 0.5 m Figure Comparison of ND-CPT measured wet density profiles with laboratory results. 67

36 Wet Density, (kn/ m 3 ) Wet Density, (kn/ m 3 ) % +5% -5% +5% 15 RI Depth (m) 16 BH 10 - direct measurement BH 10 - from water content Depth (m) 16 CPT 3 BH 11 - direct measurement BH 11 - from water content 17 RI BH 11 BH 10 CPT m 0.5 m 0.5 m Figure Comparison of ND-CPT measured wet density profiles with laboratory results. 68

37 22 ND-CPT measured (kn/m 3 ) Data's lie w ithin 5% bounds Data's lie outside 5% bounds Lab measured (kn/m 3 ) Figure 3.20 (a) ND-CPT versus Laboratory measured wet density Frequency Mean = 1.01 SD = COV = N = Correction Factor (β) Figure 3.20 (b) Frequency versus measurement bias. 69

38 R ρ = ρ t ρ t 2 (N=444, r 2 = 0.98) Density Count Ratio, Rρ Japan Soils Singapore Clay Wet Density, ρ t (t/m 3 ) Figure 3.21 (a) Improved calibration curve of ND-CPT Frequency Mean = 1.00 SD = 0.03 COV = 0.03 N = Correction Factor (β) Figure 3.21 (b) Frequency versus measurement bias 70