Baixe o app para aproveitar ainda mais
Prévia do material em texto
Arshad, M. I. et al. (2014). Ge´otechnique 64, No. 7, 551–569 [http://dx.doi.org/10.1680/geot.13.P.179] 551 Experimental study of cone penetration in silica sand using digital image correlation M. I . ARSHAD�, F. S . TEHRANI�, M. PREZZI� and R. SALGADO� The problem of cone penetration, particularly deep penetration, remains one of the most challenging in geotechnical engineering. It involves large displacements, rotations and deformation of soil elements in the path of the cone as well as complex response of the soil, including crushing and the development of large mean stresses, to the displacements imposed by the penetration process. As a result, rigorous theoretical solutions are not available for this problem, and experimental simulations of penetration provide insights that would not otherwise be available. This paper presents the results of a series of cone penetration tests performed in a half-circular chamber in sand samples prepared with three silica sands with different crushability. Cone resistance was measured, and digital images of the cone penetrating into the sand samples were acquired simultaneously during the entire penetration process. The digital image correlation (DIC) technique was then used to process these images to obtain the soil displacement field resulting from cone penetration. The results of DIC analyses and measured cone resistance suggest that the soil displacement around an advancing cone depends on the density and crushability of the sand, as well as the depth of penetration. Tests on silica sands with different degrees of crushability show that, for shallow penetration, the displacement vectors near the cone tip are essentially vertical for crushable sand, transitioning to subvertical for less crushable sands. However, for deep penetration, the displacement vectors near the cone tip are mostly vertical below the cone tip. Crushing was observed immediately below and around the cone tip for all sands tested. After passage of the cone, the crushed particles form a thin, crushed particle band of thickness equal to about 2.5D50 along the shaft, with a smaller percentage of crushed particles observed within an outer band with thickness equal to 4D50. KEYWORDS: model tests; particle crushing/crushability; piles; sands; soil/structure interaction INTRODUCTION The cone penetration test (CPT) has become a key tool in site characterisation, partly owing to its simplicity, and partly as a result of the development of cone resistance-based correlations for shallow foundation design (Schmertmann, 1970; Mayne & Poulos, 1999; Lee & Salgado, 2002; Lee et al., 2005; Foye et al., 2006; Lee et al., 2008; O’Loughlin & Lehane, 2010), pile design (Lee & Salgado, 1999; Lee et al., 2003; Jardine et al., 2005; Kolk et al., 2005; Xu et al., 2008; Foye et al., 2009; Seo et al., 2009; Niazi & Mayne, 2013) and liquefaction resistance estimation (Seed & De Alba, 1986; Stark & Olson, 1995; Salgado et al., 1997a; Robertson & Wride, 1998; Carraro et al., 2003; Idriss & Boulanger, 2006). The apparent simplicity of the CPT, how- ever, hides considerably complex mechanics (Salgado, 2013). These include possible partial drainage during pene- tration; particle size effects; the large displacements, rota- tions, shear deformations and mean stresses that develop around the cone; and the particle crushing that takes place near the cone. This complexity leads to significant chal- lenges in theoretical modelling of cone penetration. The most applicable analyses for sand have been based on cavity expansion analysis (Salgado et al., 1997b; Salgado & Ran- dolph, 2001; Salgado & Prezzi, 2007). The value of experimental modelling increases when the target problem, cone penetration in the present case, is difficult to solve theoretically. Even as attempts at a rigorous theoretical solution of the cone penetration problem gain sophistication, it will still be important to have companion experimental results that can be used to validate theoretical solutions. Image-based analysis methods can be used in laboratory investigations of the cone penetration process in particulate media. X-ray film photogrammetry (Robinsky et al., 1964) and, more recently, X-ray television (TV) imaging (Kobayashi & Fukagawa, 2003) have been used for this purpose. These techniques rely on embedded lead shot markers for tracking motion within the soil (White, 2002), and metal marker separation from the soil particles at large deformations affect the precision of the observed displacements. Additionally, the scanning time involved prevents the applicability of this tech- nique to observation of continuous penetration in large-scale models. The use of advanced non-destructive techniques, such as X-ray computed tomography (CT), in geotechnical model- ling has also been restricted to small-scale models (Otani, 2004; Morita et al., 2007; Ando` et al., 2012; Paniagua et al., 2013). Stereo photogrammetry has been demonstrated for planar displacement measurement using a false relief tech- nique (Butterfield et al., 1970), but its accuracy is dependent upon the grain size of the deforming medium and texture contrast. Similarly, laser speckle interferometry (de Pater & Nieuwenhuis, 1986) has been used to observe the displace- ment field during cone penetration. This method relies on formation of laser fringe patterns due to sand grain move- ments and requires a minimum of 1 mm cone displacement for pattern recognition. Optical analysis around penetrating probes has also been employed (Allersma, 1987) for stress visualisation in plane-strain conditions, and then extended to displacement measurement by digital imaging of the same observation plane (Dijkstra et al., 2006). Transparent soil Manuscript received 27 October 2013; revised manuscript accepted 16 May 2014. Published online ahead of print 30 June 2014. Discussion on this paper closes on 1 December 2014, for further details see p. ii.� Purdue University, West Lafayette, USA. particles have been used for measurement of displacement during penetrometer/pile penetration (Iskander et al., 2002; Liu, 2003; Lehane & Gill, 2004; Ni et al., 2010). In general, the major limitations of these methods are their relatively low resolution and accuracy in measuring displacements of gran- ular materials and their reliance on photoelastic materials and synthetic particles, which may not have properties similar to those of real soils. In the past decade, image correlation techniques, such as particle image velocimetry (PIV) and digital image correla- tion (DIC), have gained attention in experimental geomech- anics. In the DIC technique, a transparent observation window, located exactly where one of the symmetry planes of a penetrating probe or model pile with a full cross section would be (possible because the probe has a half cross section), enables the observer to study the response of soil during the penetration or loading process. After image processing, the displacement field within the soil mass is obtained at different stages of the penetration or loading process. In a pioneering work, White & Bolton (2004) used the PIV technique to study the displacement of soil surrounding a model ‘pile’ during penetration under plane-strain condi- tions. In order to simulate three-dimensional (3D) penetra- tion, Liu (2010) performed centrifuge CPTs under axisymmetric conditions combined with PIV image analysis. Centrifuge testing is similar to chamber testing with respect to the PIV/DIC application, with the advantage of continu- ous stress profiling. However, in most cases, deep penetra- tion of the cone is accompanied by intrusion of sand particles between the cone and the transparent observation window, possibly reducing the accuracy of the analysis. To reduce the sand intrusion problem at deep penetration in axisymmetric models, a coarser sand is desirable; its usein a centrifuge test may, however, affect the results because of particle size effects, which can become significant at low cone diameter-to-particle size ratios (Foray et al., 1998; Loukidis & Salgado, 2008; Salgado, 2013). This paper describes CPTs performed in a half-circular DIC chamber to study the cone penetration process in silica sand with uniform density. The DIC technique was used to compute the soil displacement at shallow and deep penetra- tion from images stored during penetration. The paper examines the insights provided by the computed displace- ment fields. MATERIALS AND METHODS Test equipment The half-circular steel chamber used in the present re- search has a diameter Dc of 1680 mm and a height of 1200 mm. Jacking and loading systems and a reaction frame enabled penetration of the model cone penetrometer (see Fig. 1) into the samples. The front wall of the half-circular DIC chamber is a reinforced poly(methyl methacrylate) (PMMA) sheet. The reinforcement comes from a steel frame that contains three observation windows, which allow ima- ging of the half-circular cone penetrometer and the sur- rounding soil medium during the penetration process. A clear, annealed glass sheet is fixed behind the PMMA to prevent scratching of the PMMA by the sand particles during testing. Calibration chamber testing offers the advantage that the initial state of the soil is known; however, for chambers of small diameter, the effect of the chamber boundaries on model piles or penetrating probes must be considered in the interpretation of results (Ghionna & Jamiolkowski, 1991; Salgado et al., 1998). The effect of chamber boundaries can be minimised by choosing an appropriate chamber-to-cone diameter ratio (Salgado, 2013). The diameter dc of the model cone penetrometer is 31.75 mm. A ratio of Dc /dc greater than 50 was considered sufficient to minimise the effect of the chamber boundaries on the tests while allowing preparation of samples that are not unmanageably large. The cone diameter-to-D50 ratio of the test sands remained greater than 20 to avoid scale effects (Gui & Bolton, 1998; Salgado, 2013). Fig. 2 shows the brass model cone penetrometer, which consists of a half-circular rod section 915 mm long, to which a half-section conical tip with a 608 apex angle is attached. To measure the cone resistance qc during testing, a compression load cell with 10 kN capacity was embedded between the cone tip and the rod (see Fig. 2). A removable jacking system with 50 kN capacity was used to push the penetrometer into the sand at a penetration (a) Half-circular penetrometer Jacking system 1200 mm 1680 mm Observation windows Top lid Sand discharge valve Soil sample depth 1000 mm (b) Fig. 1. Schematic arrangement of the half-circular DIC chamber: (a) front view; (b) side view 552 ARSHAD, TEHRANI, PREZZI AND SALGADO rate of � 1 mm/s (0.83 mm/s). A 20 kN tension–compression load cell embedded between the jack and the head of the penetrometer measured the jacking load during penetration. A roller assembly consisting of three cam followers and a carefully aligned jacking system practically eliminated intru- sion of sand between the penetrometer and the glass. Grease on the penetrometer surface in direct contact with the glass minimised friction. A specially designed, half-circular, air–rubber bladder was used to apply pressure on the top surface of the sample. A half-circular steel plate was placed below the rubber bladder to ensure uniform pressure application on the top of the sample. A reaction steel lid was bolted on the top of the chamber to constrain the air bladder. Image and data acquisition system The image acquisition system used to record the images consists of three complementary metal-oxide-semiconductor (CMOS), machine-vision digital cameras with 5 mega pixel (MP) resolution. The cameras are equipped with compatible high-resolution and low-distortion lenses of 12.5 mm fixed focal length. A frame grabber with on-board memory was used to record images during the penetration tests. The sensors used for measuring the jacking force and cone resistance during cone penetration were connected to a separate data acquisition system. Base boundary effects were monitored during testing by a load cell installed in the sample at the base of the chamber along the cone penetra- tion path. Lateral boundary effects were monitored by two miniature pressure transducers installed at and near the chamber boundary at sample mid-height. Test sands The main test sand was a coarse-grained, unground silica sand (#2Q-ROK). This sand, which was mined from the Oriskany sandstone deposits at Berkeley Springs in West Virginia, has angular particles (Fig. 3(a)) and is composed 900 800 700 600 500 400 300 200 100 0 �100 �200 V er tic al p os iti on , , re la tiv e to c on e tip : m m h �150 �100 �50 0 50 100 150 Horizontal position, : mmx Cables dc 31·75 mm� 20 kN in-line tension/compression load cell Observation window 2 Observation window 3 10 kN compression load cell Fig. 2. Configuration of half-circular model penetrometer 100 mμ (a) 100 mμ (b) 50 mμ (c) Fig. 3. Microscopic images of the test sands: (a) #2Q-ROK; (b) ASTM 20-30; (c) Ohio Gold Frac CONE PENETRATION IN SILICA SAND USING DIGITAL IMAGE CORRELATION 553 mainly of quartz (silicon dioxide (SiO2) ¼ 99%), having traces of calcite. For comparison purposes, additional tests were performed with two other silica sands: ASTM-graded 20-30 Ottawa sand (SiO2 ¼ 99.8%), henceforth referred to as ASTM 20-30, which has rounded particles (Fig. 3(b)), and Ohio Gold Frac sand (SiO2 ¼ 99.7%), which has sub- angular particles (Fig. 3(c)). These three sands are classified as poorly graded (SP) according to the Unified Classification System (ASTM D2487-11; ASTM, 2011). The maximum and minimum densities of the sands were determined based on ASTM D4253-00 (ASTM, 2006a) and ASTM D4254-00 (ASTM, 2006b). Table 1 provides the index properties of the test sands. During deep penetration, which induces large mean and shear stresses below the cone tip, particle crushing is ob- served in almost all sands (BCP Committee, 1971; White & Bolton, 2004; Otani et al., 2005; Yang et al., 2010). White & Bolton (2004) observed intense particle crushing ahead of the tip of a plane-strain penetrometer monotonically jacked into silica and carbonate sands. Yang et al. (2010) performed cyclic loading tests on model, closed-ended steel piles jacked into silica sand samples subjected to 150 kPa sur- charge. Three distinct zones were identified near the pile shaft that contained different percentages of crushed parti- cles ranging in size from 20 �m to 100 �m. In order to compare the crushability characteristics of the test sands, one-dimensional (1D) compression tests were performed on very dense samples (DR � 85%) with normal stresses of up to 10 MPa (Fig. 4). The relative breakage parameter Br, as defined by Hardin (1985), was then used to provide an assessment of the crushability characteristics of the test sands. From the results of the 1D compression tests, the #2Q-ROK sand is the most crushable, with Br ¼ 10.51%; Ohio Gold Frac sand is of intermediate crushability, with Br ¼ 2.56%; and ASTM 20-30 sand is the least crushable of the three silica sands, with Br ¼ 0.56%. To assess the possible effect of sand–glass friction on the observed displacements during cone penetration, sand–glass interface direct shear tests were performed on #2Q-ROK silica sand samples prepared at medium density (DR ¼ 65%) for confining stresses of 25, 52 and 94 kPa. Fig. 5 shows the shear stress plotted against horizontal displacement curves obtained from the interface direct shear tests;the response is almost rigid, perfectly plastic, with shear strength developing at very small displacements (of the order of 0.15–0.2 mm). The sand–glass interface friction angle is 8.58 for this silica sand, which has angular particles; smaller interface friction angles are expected for the other sands used in this study, which have rounded and sub-angular particles. This value is in line with previous work done by White (2002), who reported a sand–glass interface friction angle of 118 based on results of interface direct shear tests performed with fraction-B Leighton Buzzard silica sand. The displacement required for mobilisation of the interface friction is con- siderably less than the displacement that occurs during cone penetration. Thus, any interface friction is anticipated to have minimum or negligible effect on the motion of the sand particles near the cone and can be disregarded in the interpretation of the displacement results. Only at locations far from the penetrometer, where the displacements in the absence of the glass would be very small, or possibly in displacement reversals, would the drag from glass friction potentially be non-negligible. So the displacements obtained from the DIC technique are, in general, close estimates of the displacements that would be observed in the real pro- blem. Table 1. Index properties of test sands Sand Grain shape Silicon dioxide: % Specific gravity, Gs D50: mm Cu emax emin #2 Q-ROK Angular 99.0 2.65 0.78 1.43 0.998 0.668 ASTM 20-30� Rounded 99.8 2.65 0.72 1.20 0.742 0.502 Ohio Gold Frac Sub-angular 99.7 2.65 0.65 1.37 0.853 0.537 � Data from Cho et al. (2006). 0·45 0·50 0·55 0·60 0·65 0·70 0·75 0·1 1 10 V oi d ra tio , e Normal stress, : MPaσ �n #2Q-ROK Ohio Gold Frac ASTM 20-30 DR 85%� Significant particle crushing Some particle crushing Minimal particle crushing Fig. 4. Change of void ratio in 1D compression tests 0 4 8 12 16 20 0 2 4 6 8 S he ar s tr es s: k P a Horizontal displacement: mm DR 65%� σ � �n 25 kPa σ � �n 52 kPa σ � �n 94 kPa Fig. 5. Shear stress plotted against horizontal displacement for direct shear tests performed on #2Q-ROK silica sand–glass interface 554 ARSHAD, TEHRANI, PREZZI AND SALGADO Digital image correlation fundamentals The DIC technique involves obtaining a sequence of digital images of a body or a structure in motion and then processing these images using an image correlation scheme to produce the displacements of material elements of the body. Image correlation enables tracking of elements (groups of particles) in time. There are two families of predefined correlation criteria that are often used for this purpose (Pan et al., 2009). The first group, referred to as the cross- correlation (CC) criteria, seeks a maximum correlation of grey level intensity of pixels across sequential images. The second group, referred to as the sum-of-squared-differences (SSD) criteria, seeks the minimum difference in grey-level intensity of an image pattern (subset) in the reference and deformed/displaced images. Both families of correlation criteria are in essence complementary and can be deduced from each other (Pan et al., 2009). The choice of the matching criterion depends upon the experimental condi- tions, nature of deformations, and the quality of the speckle pattern or random texture. In general, as illustrated in Fig. 6(a), a DIC algorithm (Pan et al., 2009) involves (a) defining a virtual grid on the initial image (reference image) (b) defining the subset size and search zone in the subsequent image (deformed image) (c) interpolating the discrete grey-level intensity of the deformed image to form a continuous function (d ) establishing the correlation between image sets using a suitable criterion Reference image Subset size (L)0 x y Subset spacing f x y( , )i i Reference subset Correlation maximisation algorithm Target subset g x( ,�i y �i ) 0 x y Deformed image Displacement vector (a) ti ti�1 ti�2 ti n� �1 ti n� (b) Fig. 6. Digital image correlation: (a) procedure outline for one subset; (b) incremental correlation CONE PENETRATION IN SILICA SAND USING DIGITAL IMAGE CORRELATION 555 (e) obtaining average displacements at the intersection of measurement grids. A fundamental assumption in the DIC technique is that the image grey-level intensity pattern remains the same before and after motion but for some Gaussian noise (Sutton et al., 2009). In the case of large deformation problems, such as the cone penetration problem, some soil particles may disappear from the subset, resulting in loss of accuracy due to a mismatch between the subsets of the deformed and undeformed images. To avoid this problem, an incremental correlation scheme was adopted in this study that involves comparing an image at a given time ti with a subsequent image at the next time step ti+1 (Fig. 6(b)). The accumulated displacements at any given time were calculated by integrat- ing the incremental displacement field obtained from the incremental correlation. A disadvantage of the incremental correlation scheme is that the systematic error associated with the displacement obtained at any increment adds up. To minimise the systematic error, instead of working with the entire set of images, an optimum number of images were selected for processing. In addition, other measures were adopted to enhance the accuracy of the image correlation, such as texture enhancement, use of a low-pass filter in the analysis of the images, and use of improved correlation algorithms accounting for subset deformation. Moreover, to avoid variation in the lighting conditions during the test, a matching criterion for image correlation based on a zero- normalised sum of squared differences was employed (Sut- ton et al., 2009). The commercial digital image correlation software VIC-2D (Correlated Solutions, 2009) was used for processing the images taken during cone penetration. Other DIC parameters that affect the measurement accuracy of displacements were optimised through a parametric study; these factors include subset size, choice of correlation algo- rithm, interpolation function for subset displacement and image intensity interpolated function. Camera calibration. The displacement field obtained from image correlation is in pixels. To convert the pixel coordinates to the object space coordinates, an image transformation matrix is first obtained from camera calibration (White et al., 2003). To obtain the camera calibration parameters, a set of 45 image control points were printed on a transparent laminating film using a 1440 DPI printer. The calibration target film covers the entire inner surface of the glass that is in direct contact with the PMMA. The centroids of these control points in image space were obtained using the Matlab software geoCENTROID (White et al., 2003). These control points were then used to establish the camera calibration parameters using the camera calibration toolbox for Matlab (Heikkila & Silven, 1997). A linear fit to the image-to-object scale was chosen by minimising the standard deviation of the projection error. This procedure was followed for all tests. EXPERIMENTAL PROGRAMME Test matrix A series of ten CPTs was performed in the DIC chamber in uniform sand samples (Table 2). In total, eight tests were performed on the most crushable silica sand (#2Q-ROK) with and without surcharge. Out of these eight tests, three tests were performed without any surcharge to investigate the development of the slip mechanism at shallow penetration and the effect of stress level on the displacement field. Two tests were performed on the other two sands, ASTM 20-30 sand and Ohio Gold Frac sand. The tests are identified by acode that specifies the sand density (L denotes ‘loose’, MD denotes ‘medium dense’ and D denotes ‘dense’), surcharge (0 or 50 kPa), test number (a number from 1 to 10) and sand type (#2Q-ROK, ASTM 20-30 and Ohio Gold Frac). For example, CPTL50-T4-#2Q-ROK identifies CPT number 4, performed on a loose #2Q-ROK sand sample with a sur- charge of 50 kPa. Test procedure The medium dense and dense sand samples were prepared by air pluviation using a large pluviator (Lee et al., 2011) placed above the DIC chamber at a fixed position such that the sand drop height was always greater than the terminal sand fall height of 500 mm, determined through the calibra- tion procedure described by Rad & Tumay (1987). The target sample densities were achieved by changing the flow rate through addition or removal of a diffuser sieve. A slightly different procedure was used for preparation of loose sand samples. While carefully maintaining a drop height of 400 mm, a half-circular pluviator without diffuser sieves was placed inside the DIC chamber and raised slowly as the sample was prepared. For sample density and uniformity verification, nine miniature thin tube samplers with diameter equal to 2 in (5.08 cm) were placed at different heights within the sand sample during sample preparation and were used to determine density locally. The density variation in a sample was determined to be within 2–3%. For the tests performed with a surcharge of 50 kPa, the surcharge was applied gradually on top of the sample by inflating the half- circular, air–rubber bladder using a laboratory air-pressure line. A special cross-hair alignment device was used to align the cameras with the observation windows. The penetrometer was subsequently jacked into the soil sample at a rate of 1 mm/s (0.83 mm/s). Digital images were acquired simultaneously from three cameras during the pene- tration process at a constant frame rate of 2 frames/s. The cone tip load Qc and the jacking force Qt were measured Table 2. Test programme Test code Surcharge: kPa Average relative density DR: % dc /D50 dc /Dc CPTL0-T1-#2Q-ROK 0 45 41 53 CPTMD0-T2-#2Q-ROK 0 65 CPTD0-T3-#2Q-ROK 0 85 CPTL50-T4-#2Q-ROK 50 45 CPTL50-T5-#2Q-ROK 50 42 CPTMD50-T6-#2Q-ROK 50 63 CPTD50-T7-#2Q-ROK 50 85 CPTD50-T8-#2Q-ROK 50 82 CPTD50-T9-Ohio Gold Frac 50 87 59 53 CPTD50-T10-ASTM 20-30 50 85 43 556 ARSHAD, TEHRANI, PREZZI AND SALGADO throughout each test. Feedback from the chamber base load cell below the cone path was used to monitor base boundary effects. Negligible chamber base boundary stresses were measured for penetration depths up to 15 to 20 cone diameters, depending on sand type and density. CONE PENETRATION RESISTANCE In addition to the digital images acquired during the CPTs, cone resistance profiles were obtained from the meas- urements made by the load cell embedded inside the cone tip; these are illustrated in Fig. 7 and Fig. 8. Figure 7(a) shows cone resistance profiles for the tests performed in loose, medium dense and dense sand samples without surcharge loading (CPTL0-T1-#2Q-ROK, CPTMD0- T2-#2Q-ROK and CPTD0-T3-#2Q-ROK). Fig. 7(b) shows the qc profiles for all the CPTs performed with a surcharge of 50 kPa in the #2Q-ROK silica sand. Cone resistance increases with increasing initial sample relative density in all tests. In the loose and medium dense sand samples, cone tip resistance tends to stabilise earlier than in the dense sand sample. Fig. 8 shows the cone resistance profiles for the tests performed in dense samples (DR ¼ 82%, 85% and 87%) prepared with three different silica sands (#2Q-ROK, Ohio Gold Frac and ASTM 20-30). These results clearly show the effect of particle crushing on cone resistance, as the more crushable sand (#2Q-ROK) offers less resistance to cone penetration than the sand of intermediate crushability (Ohio Gold Frac) and the sand of least crushability (ASTM 20-30). DIC RESULTS Presentation of the results The DIC analyses of images stored for each test produced (a) incremental displacement field resulting from each 0.415 mm increment of cone penetration between two consecutive frames (b) soil displacement paths (c) post-penetration accumulated displacement field. Figure 9 shows the coordinate reference system used to report the DIC results. The cone penetration depth from the top of the sample at any given time is denoted by h�. The vertical distance of a point in the domain with respect to the cone tip is h (h ¼ 0 at the cone tip, positive above the cone tip and negative below it). Results are typically presented with h and h� normalised by the cone radius (h/rc and h�/rc). The horizon- tal distance from a point to the cone penetration axis is r, whereas the vertical distance of the point to the sample surface is z. Both r and z can be normalised with respect to the cone radius (r/rc and z/rc). Soil displacement pattern during cone penetration The deformation pattern around an advancing cone has historically been viewed through the prism of a slip mechan- ism (Salgado & Prezzi, 2007). Such a slip mechanism can be inferred from the observation of the soil displacement field in the immediate neighbourhood of the cone tip for small increments of cone penetration. Paniagua et al. (2013) used 3D X-ray CT and 3D DIC algorithms to study displace- ments and strains around a cone penetrometer pushed into silt, and observed also that there was a pattern consistent with the Salgado & Prezzi (2007) slip pattern. Fig. 10 and Fig. 11 show the evolution of the slip mechanism during incremental cone penetration. For each of the analyses shown in Fig. 10 and Fig. 11, the incremental displacement field was obtained for an incremental penetration of 2.075 mm (¼ 5 3 0.415 mm ¼ 2.075 mm increment of cone penetration between images i and (i + 5)), which corre- sponds to 0.13rc. The cone displacement in Fig. 10 and Fig. 11 is shown without magnification, but the displacement vectors are magnified by a factor of 20 to allow better visualisation of the displacement field. Fig. 10 illustrates the evolution of the slip mechanism with increasing cone pene- tration for the CPTD0-T3-#2Q-ROK test. A free surface exists at the top of the sample since no surcharge was applied in this test. Image pairs of the cone penetration at three normalised penetration depths (h�/rc ¼ 2, 6 and 22) were analysed to obtain the incremental displacement fields at each of these normalised penetration depths. As the cone 600 400 200 0 0 1 2 3 4 5 qc: MPa C on e pe ne tr a tio n de pt h, : m m z CPTL0-T1-#2Q-ROK ( 45%)DR � CPTMD0-T2-#2Q-ROK ( 65%)DR � CPTD0-T3-#2Q-ROK ( 85%)DR � 600 400 200 0 0 5 10 15 20 25 qc: MPa C on e pe ne tr a tio n de pt h, : m m z (a) (b) CPTL50-T4-#2Q-ROK ( 45%)DR � CPTL50-T5-#2Q-ROK ( 42%)DR � CPTMD50-T6-#2Q-ROK ( 63%)DR � CPTD50-T7-#2Q-ROK ( 85%)DR � CPTD50-T8-#2Q-ROK ( 82%)DR � Fig. 7. Cone resistance profiles for penetration tests in #2Q-ROK silica sand: (a) tests without surcharge; (b) tests with a surcharge of 50 kPa CONE PENETRATION IN SILICA SAND USING DIGITAL IMAGE CORRELATION 557 first enters the sample, a shallow slip surface forms, with the displacement vectors fanning out in a pattern similar to that of ‘classical’ bearing capacity failure (Salgado, 2008), with the conical tip acting as a rigid driving wedge. Figure 11 compares the displacement field around the cone during incremental penetration for three tests performed on samples of three different sands (CPTD50-T8- #2Q-ROK, CPTD50-T9-Ohio Gold Frac and CPTD50-T10- ASTM 20-30) with approximately the same density (82%, 85% and 87%) subjected to the same surcharge (50 kPa). The figure shows the incremental displacement fields for two normalised depths, h�/rc ¼ 6 and h�/rc ¼ 20, correspondingto normalised depths just prior to and after the onset of particle crushing in the soil zone below the cone (particle crushing was visible to the naked eye through the observa- tion window). The incremental displacement field for h�/ rc ¼ 6 (Figs 11(a)–11(c)) is such that, immediately below the cone tip, the displacement vectors are nearly vertical, while further away from it, the displacement vectors have a larger radial component. A transition zone is observed where the incremental displacement vectors rotate from the vertical to the radial direction, as suggested by Salgado & Prezzi (2007); this zone is more clearly observed for the least crushable sand (ASTM 20-30), for which the degree of incremental displacement vector rotation is more pro- nounced. The incremental displacement field for h�/rc ¼ 20 500 400 300 200 100 0 0 6 12 18 24 qc: MPa C on e pe ne tr a tio n de pt h, : m m z CPTD50-T9-Ohio Gold Frac ( 87%)DR � CPTD50-T10-ASTM 20-30 ( 85%)DR � CPTD50-T7-#2Q-ROK ( 85%)DR � CPTD50-T8-#2Q-ROK ( 82%)DR � Fig. 8. Comparison of cone resistances for #2Q-ROK, Ohio Gold Frac and ASTM 20-30 silica sands 0 10 20 30 N or m al is ed d ep th , / z r c 10 5 0 5 10 Normalised radial position, / , relative to cone tipr rc 10 0 �10 N or m al is ed v er tic al p os iti on , / , r el a tiv e to c on e tip h r c r z h r*/ c h r/ 0c � h h h r * Cone penetration depth from top of sample with time Vertical position relative to cone top ( is positive above tip and negative below tip) Cone radius � � �c Fig. 9. Coordinate reference system for DIC data reporting 8 6 4 2 0 6 4 2 0 2 4 6 N or m al is ed d ep th , /z r c Normalised radial position, / , relative to cone tip (a) r rc σs R c 0 kPa, 85% * 2 , #2Q-ROK � � � D h r Cone incremental displacement 0·13� rc Scale for incremental displacement vectors: 0·063rc 12 10 8 6 4 6 4 2 0 2 4 6 N or m al is ed d ep th , /z r c Normalised radial position, / , relative to cone tip (b) r rc σs R c 0 kPa, 85% * 6 , #2Q-ROK � � � D h r Cone incremental displacement 0·13� rc Scale for incremental displacement vectors: 0·063rc 28 26 24 22 20 6 4 2 0 2 4 6 N or m al is ed d ep th , /z r c Normalised radial position, / , relative to cone tip (c) r rc σs R c 0 kPa, 85% * 22 , #2Q-ROK � � � D h r Cone incremental displacement 0·13� rc Scale for incremental displacement vectors: 0·063rc Fig. 10. Evolution of slip pattern with penetration for test CPTD- T3-#2Q-ROK (�s 0 kPa, DR 85%): (a) h�/rc 2; (b) h�/rc 6; (c) h�/rc 22 558 ARSHAD, TEHRANI, PREZZI AND SALGADO (Figs 11(d)–11(f)) illustrates that the overall incremental displacement field becomes more vertical for all sands as the cone penetrates to a depth at which the sand particles surrounding the conical tip undergo crushing (i.e., incremen- tal displacement rotation is less pronounced after particle crushing becomes significant). In order to quantify the degree of incremental displacement rotation within the tran- sition zone, the direction of the average incremental displacement vectors was calculated within a subdomain of the soil domain obtained by revolution of the cross section shown in Fig. 11. This subdomain is essentially a cylinder, except for the conical space occupied by the penetrometer tip, with radius 2dc and length extending from the cone shoulder down to 1.25dc below the cone tip. The average direction of the incremental displacement vectors is calcu- lated using the product of the radius at which the displace- ment is observed and the magnitude of the displacement as a weight factor Łave ¼ Pnel i¼1 Łili2�ridrPnel i¼1 li2�ridr (1) where Łi ¼ tan�1 vi ui � � (2) 12 10 8 6 4 6 4 2 0 2 4 6 N or m al is ed d ep th , /z r c Normalised radial position, / , relative to cone tip (a) r rc σs R c 50 kPa, 85% * 6 , ASTM 20-30 � � � D h r Cone incremental displacement 0·13� rc Scale for incremental displacement vectors: 0·063rc 12 10 8 6 4 6 4 2 0 2 4 6 N or m al is ed d ep th , /z r c Normalised radial position, / , relative to cone tip (c) r rc σs R c 50 kPa, 82% * 6 , #2Q-ROK � � � D h r Cone incremental displacement 0·13� rc Scale for incremental displacement vectors: 0·063rc Direction of average displacement vector 13·2°� 12 10 8 6 4 6 4 2 0 2 4 6 N or m al is ed d ep th , /z r c Normalised radial position, / , relative to cone tip (b) r rc σs R c 50 kPa, 87% * 6 , Ohio Gold Frac � � � D h r Cone incremental displacement 0·13� rc Scale for incremental displacement vectors: 0·063rc Direction of average displacement vector 33·2°� Direction of average displacement vector 31·5°� Direction of average displacement vector 33·0°� 26 24 22 20 18 6 4 2 0 2 4 6 N or m al is ed d ep th , /z r c Normalised radial position, / , relative to cone tip (d) r rc σs R c 50 kPa, 85% * 20 , ASTM 20-30 � � � D h r Cone incremental displacement 0·13� rc Scale for incremental displacement vectors: 0·063rc Direction of average displacement vector 40·1°� 26 24 22 20 18 6 4 2 0 2 4 6 N or m al is ed d ep th , /z r c Normalised radial position, / , relative to cone tip (e) r rc σs R c 50 kPa, 87% * 20 , Ohio Gold Frac � � � D h r Cone incremental displacement 0·13� rc Scale for incremental displacement vectors: 0·063rc Direction of average displacement vector 36·8°� 26 24 22 20 18 6 4 2 0 2 4 6 N or m al is ed d ep th , /z r c Normalised radial position, / , relative to cone tip (f) r rc σs R c 50 kPa, 82% * 20 , #2Q-ROK � � � D h r Cone incremental displacement 0·13� rc Scale for incremental displacement vectors: 0·063rc Fig. 11. Soil displacement pattern for ASTM 20-30, Ohio Gold Frac and #2Q-ROK silica sands prior to (h�/rc 6) and after particle crushing (h�/rc 20): (a) ASTM 20-30 at h�/rc 6; (b) Ohio Gold Frac at h�/rc 6; (c) #2Q-ROK at h�/rc 6; (d) ASTM 20-30 at h�/rc 20; (e) Ohio Gold Frac at h�/rc 20; (f) #2Q-ROK at h�/rc 20 CONE PENETRATION IN SILICA SAND USING DIGITAL IMAGE CORRELATION 559 li ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u2i þ v2i q (3) and ui and vi are the radial and vertical displacements of every soil element i within the averaging zone, respectively, Łi is the direction of the displacement vector of element i with respect to the horizontal, li is the magnitude of the displacement vector for element i, nel is the number of soil elements within the subdomain where the average is calcu- lated, and ri is the radial distance between the centre of element i and the cone axis. Characterisation of displacement zones around the cone Four zones providing a general, qualitative sense of the displacement field and of processes taking place during cone penetration were identified based on the displacement pat- terns observed (Fig. 12(a)): zone I – a vertical compression zone below the cone; zone II – a crushed particle band along the surface of the cone tip and shaft; zone III – a transition zone where the displacement vectors rotate from the vertical/sub-vertical direction to the radial direction; and zone IV – a zone in which the direction of the displace- ments is approximately radial. Figure 12(b) shows a vector plot of the incremental displacementfield obtained when the cone tip moved down from h� ¼ 6rc (initial position) to h� ¼ 7rc (final position) for CPTD50-T8-#2Q-ROK. The four zones are sketched in the same plot. In zone I, vertical compression of the soil dominates, and the incremental displacement vectors are mostly vertical. For small penetration increments, zone I extends from the cone tip to h/rc � �3.8; for lower eleva- tions, the incremental vertical displacement reduces to less than 10% of the maximum incremental vertical displacement observed in zone I. The maximum incremental vertical displacement is observed along the cone surface at r/rc � 0.5 and h/rc ¼ 0.866. Zone II forms due to intense shearing and particle crush- ing that takes place below the cone. The particles crushed below the cone are displaced laterally as the cone advances, and stay roughly in the same place as the cone continues in its advance. The finely crushed sand particles can be ob- served coating the cone tip and the entire cone shaft in close-up views of the images captured during penetration. The maximum thickness of this zone, which was also ob- served by White & Bolton (2004) and Yang et al. (2010), is 1.5–4 D50 at r/rc � 0.5 and h/rc ¼ 0.866. Zone III exists immediately below the cone, where rota- tion of the incremental displacement field from a mainly vertical direction to the radial direction occurs, as can be seen in Fig. 12(b). The inclination of the outer boundary of the transition zone III depends on sand crushability: the more crushable the sand, the less the boundary rotates outward (away from the cone). In zone IV, the incremental soil displacement field is purely radial, resembling the field that would result from a cylindrical cavity expansion pro- cess, with minimal vertical incremental displacement ob- served. Soil displacement in each of zones I though IV can be better understood by observing the displacement paths of key soil elements (labelled E1 through E6 in Fig. 12(a)). Fig. 13 shows the evolution of the normalised vertical and radial displacement increments ˜v/rc and ˜u/rc of elements E1, E2, E3, E4, E5 and E6 shown in Fig. 12(a) due to a penetration of approximately 1rc (¼ 1.05 rc) from a depth h� ¼ 20rc. The incremental change in the displacement field was deduced from the analysis of image sequences obtained in tests CPTD50-T10-ASTM 20-30 and CPTD50-T8-#2Q- ROK. Fig. 13(a) shows the displacement path of element E1 located just below the cone tip in zone I. Element E1 experiences a maximum vertical displacement increment ˜v ¼ 0.68rc for the least crushable sand (ASTM 20-30) and ˜v ¼ 0.49rc for the most crushable sand (#2Q-ROK). The radial displacement change is negligible for both sands. It is seen in Fig. 13(a) that particle tracking for element E1 ends before cone penetration equal to rc is achieved. For element E2 next to the cone surface, maximum ˜v values of 0.34rc and 0.48rc were observed for the least and most crushable sand, respectively (Fig. 13(b)). The increase in the normalised vertical displacement increment observed for the most crushable silica sand #2Q-ROK is caused by the intense particle crushing occurring just below the cone tip. Element E2 undergoes a maximum radial displacement increment equal to 0.17rc for the ASTM 20-30 sand and to 0.13rc for the #2Q-ROK sand. Element E3 (Fig. 13(c)), located in the transition zone, experiences maximum radial and vertical displacement increments of 0.12rc and 0.10rc for ASTM 20-30 sand, and 0.12rc and 0.16rc for #2Q-ROK sand. The displacement paths of elements E4, E5 and E6, located on the transition line between zone III and zone IV, indicate clearly that the radial component of the normalised 0 1 2 3 4 5 r r/ c Symbol shows element location (a) 2 1 0 �1 �2 �3 h r/ c Interface zone Crushed particles band thickness 1·5–4� D50 II E4 IV Cavity expansion zone (radial displacement dominates) E2 E3 E5 E6 III Transition zone (radial displacement build-up) E1 I Axial compression zone (vertical displacement dominates) Contour line corresponds to displacement ratio ( / ) 1u v � N or m al is ed d ep th , /z r c σs R c 50 kPa, 82% * 7 , #2Q-ROK � � � D h r Normalised radial position, / , relative to cone tip (b) r rc Zone II 5 7 9 11 13 6 4 2 0 2 4 6 Zone I Zone III Zone IV 0·2rc Scale for incremental displacement vectors Fig. 12. Characterisation of displacement zones near cone during a penetration equal to rc: (a) displacement zones; (b) displacement field change 560 ARSHAD, TEHRANI, PREZZI AND SALGADO 0 0·2 0·4 0·6 0·8 20·0 20·4 20·8 21·2 N or m al is ed d is pl ac em en t c ha ng e, / an d / Δ Δ u r v r c c Normalised penetration depth, */ (a) h rc σs 1 c 50 kPa Element E : 0, 20·32 � � �r z r Radial (ASTM 20-30, 85%)DR � Vertical (ASTM 20-30, 85%)DR � Radial (#2Q-ROK, 82%)DR � Vertical #2Q-ROK( , 82%)DR � σs 2 c c 50 kPa Element E : 0·68 , 18·8 � � �r r z r Radial (ASTM 20-30, 85%)DR � Vertical (ASTM 20-30, 85%)DR � Radial (#2Q-ROK, 82%)DR � Vertical #2Q-ROK( , 82%)DR � 0 0·1 0·2 0·3 0·4 0·5 20·0 20·4 20·8 21·2 N or m al is ed d is pl ac em en t c ha ng e, / an d / Δ Δ u r v r c c Normalised penetration depth, */ (b) h rc σs 3 c c 50 kPa Element E : 1 , 19·2 � � �r r z r Radial (ASTM 20-30, 85%)DR � Vertical (ASTM 20-30, 85%)DR � Radial (#2Q-ROK, 82%)DR � Vertical #2Q-ROK( , 82%)DR � σs 4 c c 50 kPa Element E : 1·12 , 18·2 � � �r r z r Radial (ASTM 20-30, 85%)DR � Vertical (ASTM 20-30, 85%)DR � Radial (#2Q-ROK, 82%)DR � Vertical #2Q-ROK( , 82%)DR � 0 0·04 0·08 0·12 0·16 0·20 20·0 20·4 20·8 21·2 N or m al is ed d is pl ac em en t c ha ng e, / an d / Δ Δ u r v r c c Normalised penetration depth, */ (c) h rc 0 0·02 0·04 0·06 0·08 20·0 20·4 20·8 21·2 N or m al is ed d is pl ac em en t c ha ng e, / an d / Δ Δ u r v r c c Normalised penetration depth, */ (d) h rc σs 5 c c 50 kPa Element E : 2 , 19·2 � � �r r z r Radial (ASTM 20-30, 85%)DR � Vertical (ASTM 20-30, 85%)DR � Radial (#2Q-ROK, 82%)DR � Vertical #2Q-ROK( , 82%)DR � σs 6 c c 50 kPa Element E : 2·7 , 20 � � �r r z r Radial (ASTM 20-30, 85%)DR � Vertical (ASTM 20-30, 85%)DR � Radial (#2Q-ROK, 82%)DR � Vertical #2Q-ROK( , 82%)DR � 0 0·02 0·04 0·06 0·08 20·0 20·4 20·8 21·2 N or m al is ed d is pl ac em en t c ha ng e, / an d / Δ Δ u r v r c c Normalised penetration depth, */ (e) h rc 0 0·01 0·02 0·03 20·0 20·4 20·8 21·2 N or m al is ed d is pl ac em en t c ha ng e, / an d / Δ Δ u r v r c c Normalised penetration depth, */ (f) h rc Fig. 13. Evolution of radial and vertical displacement change for 1rc penetration in the least crushable (CPTD50-T10-ASTM 20-30) and most crushable (CPTD50-T8-#2Q-ROK) silica sands CONE PENETRATION IN SILICA SAND USING DIGITAL IMAGE CORRELATION 561 displacement increment dominates there for both sands (see Figs 13(d)–13(f)). However, the values of the radial displacement increment of elements located at larger radial offsets from the cone tip (elements at positions E5 and E6) are smaller for the most crushable silica sand than those observed for the least crushable silica sand (0.036rc and 0.017rc, as opposed to 0.067rc and 0.02rc). These results suggest that deformation gets more localised near the conein the most crushable sand. Displacement paths during continuous penetration Figure 14 shows the displacement paths of nine elements located at different radial offsets from the cone penetration axis (r/rc ¼ 0, 0.5, 1, 1.5, 2, 3, 4, 6 and 8) for test CPTD50- T8-#2Q-ROK. The displacement paths follow the coordi- nates of the centroid of each of the soil elements during cone penetration from h� ¼ 0 to h� ¼ 30rc. The character- istics of the displacement paths of each of these elements, located initially at a depth z � 18rc, depend on their offset position with respect to the cone penetrometer path. The displacement paths for soil elements A, B, C, D and E located at r/rc ¼ 0.5, 1, 1.5, 2 and 3 show that these elements move away from the cone path, except near the end of the displacement paths, when the soil elements move slightly towards the cone shaft instead (see Fig. 15). For elements F, G and H located at r/rc ¼ 4, 6 and 8, no inward displacement is observed at the end of the displacement paths. Vertical displacements, which accumulate below the cone tip, decay sharply with increasing radial distance from the cone penetration path. As can be seen in Fig. 14, the soil displacement paths radiating from the cone tip are inclined, with radial and vertical displacements decaying with increasing radial distance from the cone tip. However, the rate of decay of the radial displacement with increasing distance from the cone path is lower than that of the vertical displacement. Figure 15 shows close-up views of the displacement paths for elements C and H located at r/rc ¼ 1.5 and at r/rc ¼ 8, respectively. For element C (Fig. 15(a)), the vertical compo- nent of displacement dominates initially, but, on further cone penetration, the displacement becomes more radial and, immediately after the cone shoulder crosses the elevation of element C, bends back towards the cone shaft. Fig. 15(b) shows the displacement path of element H. The radial and vertical displacements are of approximately the same magni- tude, with no inward displacement towards the shaft taking place as the cone penetrates beyond the elevation of this element. As shown in Fig. 15(a), the motion of element C undergoes a reversal, and its inward motion after this reversal is significantly less than its outward displacement. Given that the inward motion of element C is small, friction between the glass and the sand particles may have somewhat impeded the inward motion since the motion reversal is 300 290 280 270 �20 0 20 40 60 80 100 120 140 V er tic al p os iti on : m m Horizontal position: mm h r*/ 0c � O A B C D E F G H h r*/ 30c � Thin crushed particle band 2·5� D50 End of particle tracking C on e pe ne tr a tio n pa th σs R50 kPa, 82%, #2Q-ROK� �D Vertical shearing Intense rotation Element A (0·5 )rc Element B (1·0 )rc Element C (1·5 )rc Element D (2·0 )rc Element E (3·0 )rc Element O (0 )rc Element E (4·0 )rc Element F (6·0 )rc Element G (8·0 )rc Fig. 14. Displacement paths during cone penetration from h� 0rc to h� 30rc for soil elements O, A, B, C, D, E, F, G and H 562 ARSHAD, TEHRANI, PREZZI AND SALGADO linked to unloading, which is initially approximately elastic. As mentioned earlier, the glass–soil stress–strain relationship is a stiff, linear elastic curve, which makes it possible to estimate any drag that may have been present. The inward movement of element C and others like it reflect a decrease in radial stress after the cone shoulder crosses the element elevations, and this drop in radial stress, experienced on the penetrometer shaft, implies a drop of shear stress. Figure 16 and Fig. 17 show the evolution of the normal- ised vertical and radial displacements of soil elements B, C, D and E as the cone penetrates from h� ¼ 0 to 23 and 25rc, respectively, for tests CPTL50-T4-#2Q-ROK and CPTD50- T8-#2Q-ROK. Initially, the soil elements are located at z ¼ 18rc for the loose sand sample and at z ¼ 18.5rc for the dense sample and at radial offsets of 1rc (element B), 1.5rc (element C), 2rc (element D) and 3rc (element E) from the cone penetration axis. Elements closer to the cone penetration path, such as element B, are the first elements to sense the approaching cone. This happens when the vertical distance from element B to the cone tip is equal to approximately 15 to 16rc for loose and dense sand samples, respectively. The normalised radial displacement of soil element B increases sharply when the cone tip reaches a distance of approximately 3–3.5rc from soil element B and then moves closer to it. This build up in radial stress is referred to in this paper as the cavity expansion stage (Fig. 16(a)), the start of which is labelled as point 2 in Fig. 16. Element B is displaced radially and downward as the cone passes by it. If one chooses to view the cone as stationary, then the element could be viewed as flowing around the cone face. The normalised radial soil displacement of element B peaks when the cone shoulder starts to align with it at the end of the cavity expansion process (see the peak in normalised displacements, identified as point 3 in Fig. 16(a) and Fig. 17(a)). During the cavity expansion process, particle crushing is observed near the cone tip, with the crushed particle band starting to form along the cone tip shoulder. A small reduction in normalised radial and vertical displacements takes place after the cavity expansion process (see points 3 and 4 in Fig. 16(a) and Fig. 17(a)). For the elements located farther away from the cone, the peak in normalised radial displacement is either less pronounced or no longer observed. As the cone penetration continues, further vertical shear- ing is observed above the cone only within the thin crushed particle band (zone II) along the shaft. Particle movement is very random within the crushed particle band. As observed from close-up images of the cone penetration process, for tests performed with no surcharge (such as CPTD0-T3-#2Q- ROK), vertical particle movement is more pronounced in loose sand than in dense sand prior to particle crushing. In the case of the penetration tests performed in dense sand with the 50 kPa surcharge (such as CPTD50-T8-#2Q-ROK), noticeable particle crushing takes place below the cone tip. Fig. 18 shows a close-up image of the soil in the zone (1.25rc 3 1rc) immediately next to the cone shaft for CPTD50-T8-#2Q-ROK. As shown in Fig. 18, a very thin band of thickness equal to about 2.5D50 with highly crushed particles is observed right next to the cone shaft. Next to this thin band, there is a 4D50-thick band consisting of moderately crushed sand particles. Displacement paths below the cone tip Figure 19 shows normalised radial and vertical displace- ment paths below the cone tip for h�/rc � 20 for tests performed on sand samples prepared with three different initial densities (CPTL50-T4-#2Q-ROK, CPTMD50-T6-#2Q- ROK and CPTD50-T8-#2Q-ROK). The normalised displace- ment paths are given for soil elements located at r/rc � 1, 1.5, 2, 3, 5 and 8. Both the radial and vertical normalised displacements decrease with increasing r. The maximum value of the normalised radial displacement u/rc, which is observed at an offset r/rc ¼ 1 along the shoulder of the cone tip for dense and medium dense sand (Figs 19(a) and 19(c)), is equal to 0.33, whereas u/rc ¼ 0.28 at 0.5rc below the cone tip for loose sand (Fig. 19(e)). The normalised radial displacement decays sharply with increasing vertical depth below the cone tip, becoming negligible (u/rc � 1% of the maximum normalised radial displacement observed) at h/rc equal to �10, �12 and �14 for loose, medium and dense sands, respectively (Figs 19(a), 19(c) and 19(e)). The oppo- site trend is observed in the case of the maximum normal- ised vertical displacement. The maximumnormalised vertical displacement v/rc observed at r/rc ¼ 1 at h ¼ 0 is equal to 0.8 for loose sand, as shown in Fig. 19(f), and to approximately 0.6 for both the medium dense and dense sands, as shown in Figs 19(b) and 19(d), respectively. At larger r/rc (� 1.5, 2, 3, 5 and 8) from the cone tip, the rate of decay of the normalised vertical displacement is greater 284 283 282 281 280 279 278 24 25 26 27 28 29 30 V er tic al p os iti on : m m Horizontal position: mm (a) Cone shoulder aligns with element Inward motion towards penetrometer shaft Element C (1·5 )rc σs R50 kPa, 82%, #2Q-ROK� �D 279·0 278·8 278·6 278·4 278·2 278·0 277·8 127·0 127·2 127·4 127·6 127·8 128·0 128·2 V er tic al p os iti on : m m Horizontal position: mm (b) Cone shoulder aligns with element No inward motion towards penetrometer shaft Element H (8 )rc σs R50 kPa, 82%, #2Q-ROK� �D Fig. 15. Displacement paths for: (a) element C and (b) element H CONE PENETRATION IN SILICA SAND USING DIGITAL IMAGE CORRELATION 563 for the loose sand (v/rc � 0.30, 0.19, 0.13, 0.07 and 0.02) than for the dense (v/rc � 0.34, 0.26, 0.18, 0.11 and 0.08) and medium dense sands (v/rc � 0.32, 0.23, 0.16, 0.08 and 0.04). Fig. 19 also shows that the vertical distance from the cone tip to the depth at which the normalised vertical displacement is equal to � 1% of the maximum normalised vertical displacement, is slightly greater for the dense sand (h/rc � �14) than for the medium dense (h/rc � �12) and loose sands (h/rc � �10). Post-penetration displacement field The complete displacement field at any instance of the penetration process can be obtained by combining the displacement data from all observation windows. Fig. 20 shows the normalised displacement fields obtained as the cone moved from a normalised cone penetration depth h�/rc ¼ 0 to approximately 20 for tests CPTL50-T4-#2Q-ROK and CPTD50-T8-#2Q-ROK. At h�/rc � 20, the cone tip was lo- cated at a vertical distance of 40rc from the chamber base boundary; at this position, there are no boundary effects on the displacement field. The contour plots of normalised radial and vertical displacements shown in Fig. 20 are approximately symmetric. The radial displacement extends to larger offset distances from the centre of the cone path in the dense sand than in the loose sand. Sharp vertical displacements are ob- served within a small zone in the vicinity of the cone shaft (from r ¼ 0 to r/rc , 2) for all sands. Below the cone tip, vertical displacement (v/rc) contours ranging from 0.8 to 0.05 extend vertically to h/rc � �8 and radially to r/rc � 5 for the loose sand. For the dense sand, the vertical displacement contour extends to h/rc � �12 vertically and to r/rc � 8 ra- dially. The vertical displacement contours shown in Fig. 20 are cut off at the bottom of the chart because they reach the cross bar of the calibration chamber, which prevents visualisation of displacements. �0·2 0 0·2 0·4 0·6 0 10 20 30 N or m al is ed d is pl ac em en t, / an d / u r v r c c Normalised penetration depth, */ (a) h rc #2Q-ROK 50 kPa, 45%σs R� �D Element B: 1 , 18r r z r� �c c Radial Vertical 1: Start of axial compression 2: Start of cavity expansion (radial displacement build-up) 3: End of cavity expansion 4: End of motion 3 3 4 4 2 1 �0·1 0 0·1 0·2 0·3 0·4 0 10 20 30 N or m al is ed d is pl ac em en t, / an d / u r v r c c Normalised penetration depth, */ (b) h rc #2Q-ROK 50 kPa, 45%σs R� �D Element C: 1·5 , 18r r z r� �c c Radial Vertical 1: Start of axial compression 2: Start of cavity expansion (radial displacement build-up) 3: End of cavity expansion 4: End of motion 3 3 4 4 2 1 �0·1 0 0·1 0·2 0·3 0 10 20 30 N or m al is ed d is pl ac em en t, / an d / u r v r c c Normalised penetration depth, */ (c) h rc �0·1 0 0·1 0·2 0·3 0 10 20 30 N or m al is ed d is pl ac em en t, / an d / u r v r c c Normalised penetration depth, */ (d) h rc #2Q-ROK 50 kPa, 45%σs R� �D Element D: 2 , 18r r z r� �c c Radial Vertical 1: Start of axial compression 2: Start of cavity expansion (radial displacement build-up) 3: End of cavity expansion 4: End of motion 3 3 4 4 2 1 #2Q-ROK 50 kPa, 45%σs R� �D Element E: 3 , 18r r z r� �c c Radial Vertical 1: Start of axial compression 2: Start of cavity expansion (radial displacement build-up) 3: End of cavity expansion 4: End of motion 3 3 4 4 2 1 Fig. 16. Evolution of normalised radial and vertical displacements of soil elements at different locations during cone penetration from h� 0 to 23rc for test CPTL50-T4-#2Q-ROK: (a) element B at r 1rc; (b) element C at r 1.5rc; (c) element D at r 2rc; (d) element E at r 3rc 564 ARSHAD, TEHRANI, PREZZI AND SALGADO SUMMARY AND CONCLUSIONS This paper presents the results of a series of CPTs performed in a half-circular chamber in sand samples with uniform density. Three types of silica sands, with different particle shape, size and crushability characteristics, were used to prepare samples with different densities, which were then subjected to a surcharge. Digital images of the cone moving through the sand samples were acquired simultane- ously from three cameras during the entire penetration process. The DIC technique was then used to process these images to obtain the soil displacement field resulting from both shallow and deep cone penetration. Four distinct zones were identified around the cone tip. Immediately below the cone tip, the displacement vectors were nearly vertical, while further away from it, the displace- ment vectors had a larger radial component. This is largely consistent with the coupling of a cavity expansion analysis with a separate analysis to handle the intense stress rotation observed around the cone to calculate cone resistances for a given soil state. In the zone just below the cone, the incremental displacement vectors were mostly sub-vertical 0 0·2 0·4 0·6 0 10 20 30 N or m al is ed d is pl ac em en t, / an d / u r v r c c Normalised penetration depth, */ (a) h rc #2Q-ROK 50 kPa, 82%σs R� �D Element B: 1 , 18·5r r z r� �c c Radial Vertical 1: Start of axial compression 2: Start of cavity expansion (radial displacement build-up) 3: End of cavity expansion 4: End of motion 3 3 4 4 2 1 �0·1 0 0·1 0·2 0·3 0·4 0 10 20 30 N or m al is ed d is pl ac em en t, / an d / u r v r c c Normalised penetration depth, */ (b) h rc #2Q-ROK 50 kPa, 82%σs R� �D Element C: 1·5 , 18·5r r z r� �c c Radial Vertical 1: Start of axial compression 2: Start of cavity expansion (radial displacement build-up) 3: End of cavity expansion 4: End of motion 3 3 4 2 1 0 0·1 0·2 0·3 0 10 20 30 N or m al is ed d is pl ac em en t, / an d / u r v r c c Penetration depth, */ (c) h rc 0 0·1 0·2 0·3 0 10 20 30 N or m al is ed d is pl ac em en t, / an d / u r v r c c Penetration depth, */ (d) h rc #2Q-ROK 50 kPa, 82%σs R� �D Element D: 2 , 18·5r r z r� �c c Radial Vertical 1: Start of axial compression 2: Start of cavity expansion (radial displacement build-up) 3:End of cavity expansion 4: End of motion 3 3 4 4 2 1 #2Q-ROK 50 kPa, 82%σs R� �D Element E: 3 , 18·5r r z r� �c c Radial Vertical 1: Start of axial compression 2: Start of cavity expansion (radial displacement build-up) 3: End of cavity expansion 4: End of motion 3 3 4 4 2 1 Fig. 17. Evolution of normalised radial and vertical displacements of soil elements at different locations during cone penetration from h� 0 to h� 25rc for test CPTD50-T8-#2Q-ROK: (a) element B at r rc; (b) element C at r 1.5rc; (c) element D at r 2rc; (d) element E at r 3rc Highly crushed particles Moderately crushed particles 1 ( 16 m m ) r c � 1·25 ( 20 mm)rc � σs c R50 kPa, 20 , 82%, #2Q-ROK� � �z r D 2·5 ( 2 mm) D50 � 4 ( 3 mm) D50 � Fig. 18. Close-up view of the interface zone along the penetro- meter shaft for CPTD50-T8-#2Q-ROK CONE PENETRATION IN SILICA SAND USING DIGITAL IMAGE CORRELATION 565 �16 �12 �8 �4 0 4 0·4 0·3 0·2 0·1 0 N or m al is ed v er tic al p os iti on , / , r el a tiv e to ti p h r c Normalised radial displacement, / (a) u rc h r*/ 20c � σs R50 kPa, 82% #2Q-ROK � �D r r/ 1c � r r/ 1·5c � r r/ 2c � r r/ 3c � r r/ 5c � r r/ 8c � h r*/ 20c � σs R50 kPa, 82% #2Q-ROK � �D r r/ 1c � r r/ 1·5c � r r/ 2c � r r/ 3c � r r/ 5c � r r/ 8c � �16 �12 �8 �4 0 4 0·8 0·6 0·4 0·2 0 N or m al is ed v er tic al p os iti on , / , r el a tiv e to ti p h r c Normalised vertical displacement, / (b) v rc �12 �8 �4 0 4 8 0·4 0·3 0·2 0·1 0 N or m al is ed v er tic al p os iti on , / , r el a tiv e to ti p h r c Normalised radial displacement, / (c) u rc �12 �8 �4 0 4 8 0·8 0·6 0·4 0·2 0 N or m al is ed v er tic al p os iti on , / , r el a tiv e to ti p h r c Normalised displacement, / (d) v rcvertical h r*/ 20c � σs R50 kPa, 65% #2Q-ROK � �D r r/ 1c � r r/ 1·5c � r r/ 2c � r r/ 3c � r r/ 5c � r r/ 8c � h r*/ 20c � σs R50 kPa, 65% #2Q-ROK � �D r r/ 1c � r r/ 1·5c � r r/ 2c � r r/ 3c � r r/ 5c � r r/ 8c � �12 �8 �4 0 4 0·4 0·3 0·2 0·1 0 N or m al is ed v er tic al p os iti on , / , r el a tiv e to ti p h r c Normalised radial displacement, / (e) u rc �12 �8 �4 0 4 0·8 0·6 0·4 0·2 0 N or m al is ed v er tic al p os iti on , / , r el a tiv e to ti p h r c Normalised displacement, / (f) v rcvertical h r*/ 20c � σs R50 kPa, 45% #2Q-ROK � �D r r/ 1c � r r/ 1·5c � r r/ 2c � r r/ 3c � r r/ 5c � r r/ 8c � σs R50 kPa, 45% #2Q-ROK � �D r r/ 1c � r r/ 1·5c � r r/ 2c � r r/ 3c � r r/ 5c � r r/ 8c � h r*/ 20c � Fig. 19. Displacementpathsbelowthecone tipwhenthecone isath�/rc 20: (a) radialdisplacement path for CPTD50-T8-#2Q-ROK; (b) vertical displacement path for CPTD50-T8-#2Q-ROK; (c) radial displacement path for CPTMD50-T6-#2Q-ROK; (d) vertical displacement path for CPTMD50-T6-#2Q-ROK; (e) radial displacement path for CPTL50-T4-#2Q-ROK; (f) vertical displacement path for CPTL50-T4-#2Q-ROK 566 ARSHAD, TEHRANI, PREZZI AND SALGADO for the least crushable sand, while they were mostly vertical for the most crushable sand. A very thin, crushed particle band of thickness equal to about 2.5D50 formed at the interface with the cone surface due to intense shearing and particle crushing below the cone. This thin particle band was surrounded by a 4D50-thick band consisting of moderately crushed sand particles. A transition zone, where the incre- mental displacement vectors rotated from approximately vertical to radial orientations, was also observed. A drop of radial stress was observed for elements to the side of the cone path after the cone moved down below the elevation of the elements. Crushability plays a clear role in the geometry of the displacement field. Greater crushability causes steeper displacement vectors near and below the cone and produces sharper transitions to radial displacements in the outer zone in the soil. The magnitude of the radial incremental displacement vectors in this outer zone decreased, and that �5 0 5 10 15 20 105 0 510 N or m al is ed v er tic al p os iti on , / , r el a tiv e to c on e tip h r c Normalised radial position, / , relative to cone tip (a) r rc σs R c 50 kPa, 45% * 20 , #2Q-ROK � � � D h r u r/ c 0·05 0 ·1 0· 15 0·2 0· 25 0·3 0·3 0·25 0· 2 0· 15 0· 1 0· 05 0·050·05 0·1 0·15 0·3 0·2 �5 0 5 10 15 20 105 0 510 N or m al is ed v er tic al p os iti on , / , r el a tiv e to c on e tip h r c Normalised radial position, / , relative to cone tip (b) r rc σs R c 50 kPa, 45% * 20 , #2Q-ROK � � � D h r v r/ c 0· 05 0· 1 0· 2 0·5 0· 3 0· 3 0· 5 0·2 0·1 0·05 0· 05 0·05 0·1 0·3 0·5 0·2 �5 0 5 10 15 20 105 0 510 N or m al is ed v er tic al p os iti on , / , r el a tiv e to c on e tip h r c Normalised radial position, / , relative to cone tip (c) r rc σs R c 50 kPa, 82% * 20 , #2Q-ROK � � � D h r u r/ c 0· 05 0·1 0·15 0·2 0· 25 0·35 0· 35 0·25 0· 2 0· 15 0· 1 0· 05 0·05 0·05 0·1 0·1 �5 0 5 10 15 20 105 0 510 N or m al is ed v er tic al p os iti on , / , r el a tiv e to c on e tip h r c Normalised radial position, / , relative to cone tip (d) r rc σs R c 50 kPa, 82% * 20 , #2Q-ROK � � � D h r v r/ c 0· 05 0· 1 0 ·2 0·5 0· 3 0·3 0·5 0·2 0·1 0·05 0· 05 0·05 0·1 0·3 0·5 0·2 0·8 Fig. 20. Normalised radial andvertical displacement afterconepenetration to approximatelyh�/rc 20: (a) radial displacement for CPTL50-T4-#2Q-ROK; (b) vertical displacement for CPTL50-T4-#2Q-ROK; (c) radial displacement for CPTD50-T8-#2Q-ROK; (d) vertical displacement for CPTD50-T8-#2Q-ROK CONE PENETRATION IN SILICA SAND USING DIGITAL IMAGE CORRELATION 567 of the vertical incremental displacement vectors near the cone increased with increasing particle crushing. This sug- gests that deformation becomes more localised near the cone in crushable sand. At shallow penetration, these differences are sharp for sands with different crushability; for deep penetration, every sand crushes, and the differences are less pronounced. ACKNOWLEDGEMENT This material is based upon work supported by the National Science Foundation under grant no. 0969949. The authors are very grateful for this support. NOTATION Br Hardin’s relative breakage parameter Cu coefficient of uniformity Dc diameter of DIC calibration chamber DR relative density of the sand D50 particle size larger than exactly 50% of soil particles (by weight) dc diameter of model cone penetrometer emax maximum void ratio emin minimum void ratio Gs specific gravity h vertical distance of point in domain with respect to cone tip h� cone penetration depth measured from top of the sample li magnitude of displacement vector for element i nel number of soil elements within subdomain where average direction of incremental displacement vectors is calculatedQc load measured at cone penetrometer tip Qt jacking force qc cone (tip) resistance r horizontal distance from a point to the cone penetration axis rc radius of cone penetrometer ri radial distance from centre of element i to cone penetration axis ui radial displacement vi vertical displacement z vertical distance from surface of sample to a point Łave average direction of incremental displacement vectors Łi direction of displacement vector of element i � 9n normal effective stress �s applied surcharge REFERENCES Allersma, H. G. B. (1987). Optical analysis of stress and strain in photoelastic particle assemblies. Delft, the Netherlands: Delft University of Technology. Ando`, E., Be´suelle, P., Hall, S. A., Viggiani, G. & Desrues, J. (2012). Experimental micromechanics: grain-scale observation of sand deformation. Ge´otechnique Lett. 2, July–September, 107–112. ASTM (2006a). D4253-00: Standard test methods for maximum index density and unit weight of soils using a vibratory table. ASTM International, West Conshohocken, PA, USA. ASTM (2006b). D4254-00: Standard test methods for minimum index density and unit weight of soils and calculation of relative density. ASTM International, West Conshohocken, PA, USA. ASTM (2011). D2487-11: Standard practice for classification of soils for engineering purposes (Unified Soil Classification Sys- tem). ASTM International, West Conshohocken, PA, USA. BCP Committee (1971). Field tests on piles in sand. Soils Found. 11, No. 2, 29–49. Butterfield, R., Harkness, R. M. & Andrawes, K. Z. (1970). A stereo-photogrammetric method for measuring displacement fields. Ge´otechnique 20, No. 3, 308–314, http://dx.doi.org/ 10.1680/geot.1970.20.3.308. Carraro, J. A. H., Bandini, P. & Salgado, R. (2003). Liquefaction resistance of clean and nonplastic silty sands based on cone penetration resistance. J. Geotech. Geoenviron. Engng 129, No. 11, 965–976. Cho, G., Dodds, J. & Santamarina, J. C. (2006). Particle shape effects on packing density, stiffness and strength: natural and crushed sands. J. Geotech. Geoenv. Engng, ASCE 132, No. 5, 591–602. Correlated Solutions (2009). VIC-2D and VIC-3D. See http:// www.correlatedsolutions.com (accessed 15/02/2014). de Pater, C. J. & Nieuwenhuis, J. D. (1986). Method for measuring the deformations of a sand surface. Ge´otechnique 36, No. 4, 581–585, http://dx.doi.org/10.1680/geot.1986.36.4.581. Dijkstra, J., Broere, W. & van Tol, A. F. (2006). Experimental investigation into the stress and strain development around a displacement pile. In Proceedings of the 10th international conference on piling and deep foundations (eds J. Lindberg, M. Bottiau and A. F. van Tol), pp. 252–259. Hawthorne, NJ, USA: Deep Foundations Institute. Foray, P., Balachowsky, L. & Rault, G. (1998). Scale effect in shaft friction due to the localization of deformations. In Proceedings of the international conference centrifuge 98 (eds T. Kimura, O. Kusakabe and J. Takemura), pp. 211–216. Rotterdam, the Neth- erlands: Balkema. Foye, K. C., Salgado, R. & Scott, B. (2006). Resistance factors for use in shallow foundation LRFD. J. Geotech. Geoenviron. Engng 132, No. 9, 1208–1218. Foye, K. C., Abou-Jaoude, G., Prezzi, M. & Salgado, R. (2009). Resistance factors for use in load and resistance factor design of driven pipe piles in sands. J. Geotech. Geoenviron. Engng 135, No. 1, 1–13. Ghionna, V. N. & Jamiolkowski, M. (1991). A critical appraisal of calibration chamber testing of sands. In Proceedings of 1st international symposium on calibration chamber testing, pp. 13– 39. New York, NY, USA: Elsevier. Gui, M. W. & Bolton, M. D. (1998). Geometry and scale effects in CPT and pile design. In Proceedings of 1st international con- ference on site characterization, ISC’98, Atlanta (eds P. W. Mayne and P. K. Robertson), pp. 1063–1068. Rotterdam, the Netherlands: Balkema. Hardin, B. (1985). Crushing of soil particles. J. Geotech. Engng 111, No. 10, 1177–1192. Heikkila, J. & Silven, O. (1997). A four-step camera calibration procedure with implicit image correction. Proceedings of IEEE Computer Society conference on computer vision and pattern recognition, pp. 1106–1112. New York, NY, USA: IEEE Com- puter Society, IEEE. Idriss, I. M. & Boulanger, R. W. (2006). Semi-empirical procedures for evaluating liquefaction potential during earthquakes. Soil Dynam. Earthquake Engng 26, No. 2–4, 115–130. Iskander, M. G., Samer, S. & Liu, J. (2002). Optical measurement of deformation using transparent silica gel to model sand. Int. J. Phys. Modelling Geotech. 2, No. 4, 13–26. Jardine, F. M., Chow, F. C., Overy, R. F. & Standing, J. R. (2005). ICP design methods for driven piles in sands and clays. London, UK: Thomas Telford. Kobayashi, T. & Fukagawa, R. (2003). Characterization of deformation process of CPT using X-ray TV imaging technique. In Deformation characteristics of geomaterials (eds H. di Benedetto, T. Doanh, H. Geoffroy and C. Sauze´at), pp. 43–47. Lisse, the Netherlands: Balkema. Kolk, H. J., Baaijens, A. E. & Senders, M. (2005). Design criteria for pipe piles in silica sands. Proceedings of the 1st interna- tional symposium on frontiers in offshore geotechnics, Perth, Australia, pp. 711–716. London, UK: Taylor and Francis. Lee, J. & Salgado, R. (1999). Determination of pile base resistance in sands. J. Geotech. Geoenviron. Engng 125, No. 8, 673–683. Lee, J. & Salgado, R. (2002). Estimation of footing settlement in sand. Int. J. Geomech. 2, No. 1, 1–28. Lee, J., Salgado, R. & Paik, K. (2003). Estimation of load capacity of pipe piles in sand based on cone penetration test results. J. Geotech. Geoenviron. Engng 129, No. 5, 391–403. Lee, J., Salgado, R. & Kim, S. (2005). Bearing capacity of circular footings under surcharge using state-dependent finite element analysis. Comput. Geotech. 32, No. 6, 445–457. Lee, J., Eun, J., Prezzi, M. & Salgado, R. (2008). Strain influence diagrams for settlement estimation of both isolated and multiple footings in sand. J. Geotech. Geoenviron. Engng 134, No. 4, 417–427. Lee, J., Prezzi, M. & Salgado, R. (2011). Experimental investigation 568 ARSHAD, TEHRANI, PREZZI AND SALGADO of the combined load response of model piles driven in sand. Geotech. Testing J. 34, No. 6, 1–15. Lehane, B. M. & Gill, D. R. (2004). Displacement fields induced by penetrometer installation in an artificial soil. Int. J. Phys. Modelling Geotech. 4, No. 1, 25–36. Liu, J. (2003). Visualization of three-dimensional deformations using transparent soil models. PhD thesis, Polytechnic Institute of New York University, New York, NY, USA. Liu, W. (2010). Axisymmetric centrifuge modelling of deep penetra- tion in sand. PhD thesis, University of Nottingham, Nottingham, UK. Loukidis, D. & Salgado, R. (2008). Analysis of the shaft resistance of non-displacement piles in sand. Ge´otechnique 58, No. 4, 283–296, http://dx.doi.org/10.1680/geot.2008.58.4.283. Mayne, P. W. & Poulos, H. G. (1999). Approximate displacement influence factors for elastic shallow foundations. J. Geotech. Geoenviron. Engng 125, No. 6, 453–460. Morita, K., Otani, J., Mukunoki, T., Hironaka, J. & Pham, K. D. (2007). Evaluation of vertical and lateral bearing capacity mechanisms of pile foundations using X-ray CT. In Proceedings of international workshop on recent advances of deep founda- tions (IWDPF07), Yokosuka, Japan (eds Y. Kikuchi, M. Kimura, J. Otani and Y. Morikawa), pp. 217–223. Leiden, the Nether- lands: Taylor and Francis/Balkema. Ni, Q., Hird, C. C. & Guymer, I. (2010). Physical modelling of pile penetration in clay using transparent soil and particle image velocimetry. Ge´otechnique 60, No. 2, 121–132, http://dx.doi.org/ 10.1680/geot.8.P.052. Niazi, F. S. & Mayne, P. W. (2013). Cone penetration test based direct methods for evaluating static axial capacity of single piles.
Compartilhar